From b5b92e3ef0c5a79502acc8b56e7a29571e791b90 Mon Sep 17 00:00:00 2001
From: Chris Russell <chris.russell@oii.ox.ac.uk>
Date: Wed, 19 Jun 2024 23:17:19 +0100
Subject: [PATCH] rebuild notebooks

---
 examples/adult_fairlearn_comparision.ipynb    |  657 ++++---
 examples/high-dim_fairlearn_comparision.ipynb |  317 +--
 .../multi_group_fairlearn_comparision.ipynb   |  325 +---
 examples/quickstart_xgboost.ipynb             | 1732 ++++++++---------
 4 files changed, 1388 insertions(+), 1643 deletions(-)

diff --git a/examples/adult_fairlearn_comparision.ipynb b/examples/adult_fairlearn_comparision.ipynb
index f85fa8d..c00162c 100644
--- a/examples/adult_fairlearn_comparision.ipynb
+++ b/examples/adult_fairlearn_comparision.ipynb
@@ -163,43 +163,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.187550</td>\n",
-       "      <td>0.160686</td>\n",
+       "      <td>0.200729</td>\n",
+       "      <td>0.161794</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.114406</td>\n",
-       "      <td>0.128911</td>\n",
+       "      <td>0.095843</td>\n",
+       "      <td>0.095927</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.061973</td>\n",
-       "      <td>0.014374</td>\n",
+       "      <td>0.091404</td>\n",
+       "      <td>0.021044</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.061973</td>\n",
-       "      <td>0.014374</td>\n",
+       "      <td>0.091404</td>\n",
+       "      <td>0.021044</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.081831</td>\n",
-       "      <td>0.049741</td>\n",
+       "      <td>0.101773</td>\n",
+       "      <td>0.055598</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.111037</td>\n",
-       "      <td>0.125743</td>\n",
+       "      <td>0.099629</td>\n",
+       "      <td>0.111880</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.121397</td>\n",
-       "      <td>0.123448</td>\n",
+       "      <td>0.125470</td>\n",
+       "      <td>0.134539</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.247530</td>\n",
-       "      <td>0.523846</td>\n",
+       "      <td>0.012277</td>\n",
+       "      <td>0.281679</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -207,14 +207,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.187550  0.160686\n",
-       "Predictive Parity                                0.114406  0.128911\n",
-       "Equal Opportunity                                0.061973  0.014374\n",
-       "Average Group Difference in False Negative Rate  0.061973  0.014374\n",
-       "Equalized Odds                                   0.081831  0.049741\n",
-       "Conditional Use Accuracy                         0.111037  0.125743\n",
-       "Average Group Difference in Accuracy             0.121397  0.123448\n",
-       "Treatment Equality                               0.247530  0.523846"
+       "Statistical Parity                               0.200729  0.161794\n",
+       "Predictive Parity                                0.095843  0.095927\n",
+       "Equal Opportunity                                0.091404  0.021044\n",
+       "Average Group Difference in False Negative Rate  0.091404  0.021044\n",
+       "Equalized Odds                                   0.101773  0.055598\n",
+       "Conditional Use Accuracy                         0.099629  0.111880\n",
+       "Average Group Difference in Accuracy             0.125470  0.134539\n",
+       "Treatment Equality                               0.012277  0.281679"
       ]
      },
      "execution_count": 5,
@@ -273,43 +273,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.184112</td>\n",
-       "      <td>0.157874</td>\n",
+       "      <td>0.192595</td>\n",
+       "      <td>0.148528</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.115583</td>\n",
-       "      <td>0.128469</td>\n",
+       "      <td>0.088080</td>\n",
+       "      <td>0.088552</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.063914</td>\n",
-       "      <td>0.016894</td>\n",
+       "      <td>0.061808</td>\n",
+       "      <td>0.014514</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.063914</td>\n",
-       "      <td>0.016894</td>\n",
+       "      <td>0.061808</td>\n",
+       "      <td>0.014514</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.081155</td>\n",
-       "      <td>0.049132</td>\n",
+       "      <td>0.085363</td>\n",
+       "      <td>0.049967</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.112489</td>\n",
-       "      <td>0.125768</td>\n",
+       "      <td>0.099167</td>\n",
+       "      <td>0.113456</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.121499</td>\n",
-       "      <td>0.122951</td>\n",
+       "      <td>0.129501</td>\n",
+       "      <td>0.141496</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.232591</td>\n",
-       "      <td>0.476311</td>\n",
+       "      <td>0.114719</td>\n",
+       "      <td>0.368360</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -317,14 +317,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.184112  0.157874\n",
-       "Predictive Parity                                0.115583  0.128469\n",
-       "Equal Opportunity                                0.063914  0.016894\n",
-       "Average Group Difference in False Negative Rate  0.063914  0.016894\n",
-       "Equalized Odds                                   0.081155  0.049132\n",
-       "Conditional Use Accuracy                         0.112489  0.125768\n",
-       "Average Group Difference in Accuracy             0.121499  0.122951\n",
-       "Treatment Equality                               0.232591  0.476311"
+       "Statistical Parity                               0.192595  0.148528\n",
+       "Predictive Parity                                0.088080  0.088552\n",
+       "Equal Opportunity                                0.061808  0.014514\n",
+       "Average Group Difference in False Negative Rate  0.061808  0.014514\n",
+       "Equalized Odds                                   0.085363  0.049967\n",
+       "Conditional Use Accuracy                         0.099167  0.113456\n",
+       "Average Group Difference in Accuracy             0.129501  0.141496\n",
+       "Treatment Equality                               0.114719  0.368360"
       ]
      },
      "execution_count": 6,
@@ -383,38 +383,38 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.810401</td>\n",
-       "      <td>0.806552</td>\n",
+       "      <td>0.811384</td>\n",
+       "      <td>0.803767</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.747187</td>\n",
-       "      <td>0.732342</td>\n",
+       "      <td>0.748654</td>\n",
+       "      <td>0.715969</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.612423</td>\n",
-       "      <td>0.593460</td>\n",
+       "      <td>0.614561</td>\n",
+       "      <td>0.571837</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.487204</td>\n",
-       "      <td>0.466558</td>\n",
+       "      <td>0.489989</td>\n",
+       "      <td>0.445613</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.599476</td>\n",
-       "      <td>0.596953</td>\n",
+       "      <td>0.601376</td>\n",
+       "      <td>0.598355</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.625941</td>\n",
-       "      <td>0.590007</td>\n",
+       "      <td>0.628337</td>\n",
+       "      <td>0.547570</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.747187</td>\n",
-       "      <td>0.703610</td>\n",
+       "      <td>0.748654</td>\n",
+       "      <td>0.732474</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -422,13 +422,13 @@
       ],
       "text/plain": [
        "                   original   updated\n",
-       "Accuracy           0.810401  0.806552\n",
-       "Balanced Accuracy  0.747187  0.732342\n",
-       "F1 score           0.612423  0.593460\n",
-       "MCC                0.487204  0.466558\n",
-       "Precision          0.599476  0.596953\n",
-       "Recall             0.625941  0.590007\n",
-       "ROC AUC            0.747187  0.703610"
+       "Accuracy           0.811384  0.803767\n",
+       "Balanced Accuracy  0.748654  0.715969\n",
+       "F1 score           0.614561  0.571837\n",
+       "MCC                0.489989  0.445613\n",
+       "Precision          0.601376  0.598355\n",
+       "Recall             0.628337  0.547570\n",
+       "ROC AUC            0.748654  0.732474"
       ]
      },
      "execution_count": 7,
@@ -487,38 +487,38 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.811318</td>\n",
-       "      <td>0.805995</td>\n",
+       "      <td>0.810663</td>\n",
+       "      <td>0.802883</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.744150</td>\n",
-       "      <td>0.725404</td>\n",
+       "      <td>0.744658</td>\n",
+       "      <td>0.713858</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.609492</td>\n",
-       "      <td>0.584750</td>\n",
+       "      <td>0.609723</td>\n",
+       "      <td>0.568715</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.485152</td>\n",
-       "      <td>0.458510</td>\n",
+       "      <td>0.484844</td>\n",
+       "      <td>0.442154</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.603761</td>\n",
-       "      <td>0.599353</td>\n",
+       "      <td>0.601599</td>\n",
+       "      <td>0.596841</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.615332</td>\n",
-       "      <td>0.570842</td>\n",
+       "      <td>0.618070</td>\n",
+       "      <td>0.543121</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.744109</td>\n",
-       "      <td>0.699059</td>\n",
+       "      <td>0.744658</td>\n",
+       "      <td>0.727113</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -526,13 +526,13 @@
       ],
       "text/plain": [
        "                   original   updated\n",
-       "Accuracy           0.811318  0.805995\n",
-       "Balanced Accuracy  0.744150  0.725404\n",
-       "F1 score           0.609492  0.584750\n",
-       "MCC                0.485152  0.458510\n",
-       "Precision          0.603761  0.599353\n",
-       "Recall             0.615332  0.570842\n",
-       "ROC AUC            0.744109  0.699059"
+       "Accuracy           0.810663  0.802883\n",
+       "Balanced Accuracy  0.744658  0.713858\n",
+       "F1 score           0.609723  0.568715\n",
+       "MCC                0.484844  0.442154\n",
+       "Precision          0.601599  0.596841\n",
+       "Recall             0.618070  0.543121\n",
+       "ROC AUC            0.744658  0.727113"
       ]
      },
      "execution_count": 8,
@@ -998,16 +998,13 @@
        "  /* fitted */\n",
        "  background-color: var(--sklearn-color-fitted-level-3);\n",
        "}\n",
-       "</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x30d1093c0&gt;,\n",
-       "                      estimator=DecisionTreeClassifier(),\n",
-       "                      nu=2.8954273085152237e-05)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" ><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;ExponentiatedGradient<span class=\"sk-estimator-doc-link fitted\">i<span>Fitted</span></span></label><div class=\"sk-toggleable__content fitted\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x30d1093c0&gt;,\n",
-       "                      estimator=DecisionTreeClassifier(),\n",
-       "                      nu=2.8954273085152237e-05)</pre></div> </div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-2\" type=\"checkbox\" ><label for=\"sk-estimator-id-2\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">estimator: DecisionTreeClassifier</label><div class=\"sk-toggleable__content fitted\"><pre>DecisionTreeClassifier()</pre></div> </div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-3\" type=\"checkbox\" ><label for=\"sk-estimator-id-3\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;DecisionTreeClassifier<a class=\"sk-estimator-doc-link fitted\" rel=\"noreferrer\" target=\"_blank\" href=\"https://scikit-learn.org/1.4/modules/generated/sklearn.tree.DecisionTreeClassifier.html\">?<span>Documentation for DecisionTreeClassifier</span></a></label><div class=\"sk-toggleable__content fitted\"><pre>DecisionTreeClassifier()</pre></div> </div></div></div></div></div></div></div></div></div>"
+       "</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x1649162c0&gt;,\n",
+       "                      estimator=DecisionTreeClassifier(), nu=0.0)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" ><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;ExponentiatedGradient<span class=\"sk-estimator-doc-link fitted\">i<span>Fitted</span></span></label><div class=\"sk-toggleable__content fitted\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x1649162c0&gt;,\n",
+       "                      estimator=DecisionTreeClassifier(), nu=0.0)</pre></div> </div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-2\" type=\"checkbox\" ><label for=\"sk-estimator-id-2\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">estimator: DecisionTreeClassifier</label><div class=\"sk-toggleable__content fitted\"><pre>DecisionTreeClassifier()</pre></div> </div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-3\" type=\"checkbox\" ><label for=\"sk-estimator-id-3\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;DecisionTreeClassifier<a class=\"sk-estimator-doc-link fitted\" rel=\"noreferrer\" target=\"_blank\" href=\"https://scikit-learn.org/1.4/modules/generated/sklearn.tree.DecisionTreeClassifier.html\">?<span>Documentation for DecisionTreeClassifier</span></a></label><div class=\"sk-toggleable__content fitted\"><pre>DecisionTreeClassifier()</pre></div> </div></div></div></div></div></div></div></div></div>"
       ],
       "text/plain": [
-       "ExponentiatedGradient(constraints=<fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x30d1093c0>,\n",
-       "                      estimator=DecisionTreeClassifier(),\n",
-       "                      nu=2.8954273085152237e-05)"
+       "ExponentiatedGradient(constraints=<fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x1649162c0>,\n",
+       "                      estimator=DecisionTreeClassifier(), nu=0.0)"
       ]
      },
      "execution_count": 9,
@@ -1068,78 +1065,78 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.999918</td>\n",
-       "      <td>0.810990</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.808697</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.999888</td>\n",
-       "      <td>0.742996</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.741724</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.608016</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.605405</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.999775</td>\n",
-       "      <td>0.483521</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.479245</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.603506</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.597732</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.612594</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.613279</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.999888</td>\n",
-       "      <td>0.742996</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.741724</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.194762</td>\n",
-       "      <td>0.180055</td>\n",
+       "      <td>0.194578</td>\n",
+       "      <td>0.197648</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.000202</td>\n",
-       "      <td>0.131554</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.070381</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.001131</td>\n",
-       "      <td>0.068685</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.064160</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.001131</td>\n",
-       "      <td>0.068685</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.064160</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.000610</td>\n",
-       "      <td>0.080923</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.090139</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.000170</td>\n",
-       "      <td>0.120165</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.091150</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.000062</td>\n",
-       "      <td>0.117185</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.135767</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>1.000000</td>\n",
-       "      <td>0.284807</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.024078</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1147,21 +1144,21 @@
       ],
       "text/plain": [
        "                                                    train      test\n",
-       "Accuracy                                         0.999918  0.810990\n",
-       "Balanced Accuracy                                0.999888  0.742996\n",
-       "F1 score                                         0.999829  0.608016\n",
-       "MCC                                              0.999775  0.483521\n",
-       "Precision                                        0.999829  0.603506\n",
-       "Recall                                           0.999829  0.612594\n",
-       "ROC AUC                                          0.999888  0.742996\n",
-       "Statistical Parity                               0.194762  0.180055\n",
-       "Predictive Parity                                0.000202  0.131554\n",
-       "Equal Opportunity                                0.001131  0.068685\n",
-       "Average Group Difference in False Negative Rate  0.001131  0.068685\n",
-       "Equalized Odds                                   0.000610  0.080923\n",
-       "Conditional Use Accuracy                         0.000170  0.120165\n",
-       "Average Group Difference in Accuracy             0.000062  0.117185\n",
-       "Treatment Equality                               1.000000  0.284807"
+       "Accuracy                                         1.000000  0.808697\n",
+       "Balanced Accuracy                                1.000000  0.741724\n",
+       "F1 score                                         1.000000  0.605405\n",
+       "MCC                                              1.000000  0.479245\n",
+       "Precision                                        1.000000  0.597732\n",
+       "Recall                                           1.000000  0.613279\n",
+       "ROC AUC                                          1.000000  0.741724\n",
+       "Statistical Parity                               0.194578  0.197648\n",
+       "Predictive Parity                                0.000000  0.070381\n",
+       "Equal Opportunity                                0.000000  0.064160\n",
+       "Average Group Difference in False Negative Rate  0.000000  0.064160\n",
+       "Equalized Odds                                   0.000000  0.090139\n",
+       "Conditional Use Accuracy                         0.000000  0.091150\n",
+       "Average Group Difference in Accuracy             0.000000  0.135767\n",
+       "Treatment Equality                               0.000000  0.024078"
       ]
      },
      "execution_count": 10,
@@ -1172,7 +1169,8 @@
    "source": [
     "def eval(train, classifier=mitagator):\n",
     "    return pd.concat((performance.evaluate(train['target'], classifier.predict(train['data'])),\n",
-    "                      performance.evaluate_fairness(train['target'], classifier.predict(train['data']), train['groups'])),axis=0)\n",
+    "                      performance.evaluate_fairness(train['target'], classifier.predict(train['data']), \n",
+    "                                                    train['data'][train['groups']])),axis=0)\n",
     "\n",
     "out = pd.concat((eval(train), eval(test)), axis=1)\n",
     "out.columns = ['train', 'test']\n",
@@ -1226,78 +1224,78 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.999918</td>\n",
-       "      <td>0.811318</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.810663</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.744150</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.744658</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.609492</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.609723</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.999775</td>\n",
-       "      <td>0.485152</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.484844</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.603761</td>\n",
+       "      <td>0.601599</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.999658</td>\n",
-       "      <td>0.615332</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.618070</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.744150</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.744658</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.194640</td>\n",
-       "      <td>0.184112</td>\n",
+       "      <td>0.194578</td>\n",
+       "      <td>0.192595</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.115583</td>\n",
+       "      <td>0.088080</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.000930</td>\n",
-       "      <td>0.063914</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.061808</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.000930</td>\n",
-       "      <td>0.063914</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.061808</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.000465</td>\n",
-       "      <td>0.081155</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.085363</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.000025</td>\n",
-       "      <td>0.112489</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.099167</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.000062</td>\n",
-       "      <td>0.121499</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.129501</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.232591</td>\n",
+       "      <td>0.114719</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1305,21 +1303,21 @@
       ],
       "text/plain": [
        "                                                    train      test\n",
-       "Accuracy                                         0.999918  0.811318\n",
-       "Balanced Accuracy                                0.999829  0.744150\n",
-       "F1 score                                         0.999829  0.609492\n",
-       "MCC                                              0.999775  0.485152\n",
-       "Precision                                        1.000000  0.603761\n",
-       "Recall                                           0.999658  0.615332\n",
-       "ROC AUC                                          0.999829  0.744150\n",
-       "Statistical Parity                               0.194640  0.184112\n",
-       "Predictive Parity                                0.000000  0.115583\n",
-       "Equal Opportunity                                0.000930  0.063914\n",
-       "Average Group Difference in False Negative Rate  0.000930  0.063914\n",
-       "Equalized Odds                                   0.000465  0.081155\n",
-       "Conditional Use Accuracy                         0.000025  0.112489\n",
-       "Average Group Difference in Accuracy             0.000062  0.121499\n",
-       "Treatment Equality                               0.000000  0.232591"
+       "Accuracy                                         1.000000  0.810663\n",
+       "Balanced Accuracy                                1.000000  0.744658\n",
+       "F1 score                                         1.000000  0.609723\n",
+       "MCC                                              1.000000  0.484844\n",
+       "Precision                                        1.000000  0.601599\n",
+       "Recall                                           1.000000  0.618070\n",
+       "ROC AUC                                          1.000000  0.744658\n",
+       "Statistical Parity                               0.194578  0.192595\n",
+       "Predictive Parity                                0.000000  0.088080\n",
+       "Equal Opportunity                                0.000000  0.061808\n",
+       "Average Group Difference in False Negative Rate  0.000000  0.061808\n",
+       "Equalized Odds                                   0.000000  0.085363\n",
+       "Conditional Use Accuracy                         0.000000  0.099167\n",
+       "Average Group Difference in Accuracy             0.000000  0.129501\n",
+       "Treatment Equality                               0.000000  0.114719"
       ]
      },
      "execution_count": 11,
@@ -1380,43 +1378,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.177003</td>\n",
-       "      <td>0.141355</td>\n",
+       "      <td>0.184390</td>\n",
+       "      <td>0.166026</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.007289</td>\n",
-       "      <td>0.083214</td>\n",
+       "      <td>0.019392</td>\n",
+       "      <td>0.061594</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.090675</td>\n",
-       "      <td>0.002777</td>\n",
+       "      <td>0.106631</td>\n",
+       "      <td>0.012080</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.090675</td>\n",
-       "      <td>0.002777</td>\n",
+       "      <td>0.106631</td>\n",
+       "      <td>0.012080</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.079664</td>\n",
-       "      <td>0.022729</td>\n",
+       "      <td>0.092660</td>\n",
+       "      <td>0.039273</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.050883</td>\n",
-       "      <td>0.096031</td>\n",
+       "      <td>0.057711</td>\n",
+       "      <td>0.081734</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.103941</td>\n",
-       "      <td>0.099758</td>\n",
+       "      <td>0.111285</td>\n",
+       "      <td>0.105621</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.165954</td>\n",
-       "      <td>0.228167</td>\n",
+       "      <td>0.256477</td>\n",
+       "      <td>0.186913</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1424,14 +1422,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.177003  0.141355\n",
-       "Predictive Parity                                0.007289  0.083214\n",
-       "Equal Opportunity                                0.090675  0.002777\n",
-       "Average Group Difference in False Negative Rate  0.090675  0.002777\n",
-       "Equalized Odds                                   0.079664  0.022729\n",
-       "Conditional Use Accuracy                         0.050883  0.096031\n",
-       "Average Group Difference in Accuracy             0.103941  0.099758\n",
-       "Treatment Equality                               0.165954  0.228167"
+       "Statistical Parity                               0.184390  0.166026\n",
+       "Predictive Parity                                0.019392  0.061594\n",
+       "Equal Opportunity                                0.106631  0.012080\n",
+       "Average Group Difference in False Negative Rate  0.106631  0.012080\n",
+       "Equalized Odds                                   0.092660  0.039273\n",
+       "Conditional Use Accuracy                         0.057711  0.081734\n",
+       "Average Group Difference in Accuracy             0.111285  0.105621\n",
+       "Treatment Equality                               0.256477  0.186913"
       ]
      },
      "execution_count": 12,
@@ -1483,43 +1481,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.177476</td>\n",
-       "      <td>0.143325</td>\n",
+       "      <td>0.184497</td>\n",
+       "      <td>0.167118</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.010084</td>\n",
-       "      <td>0.070287</td>\n",
+       "      <td>0.025649</td>\n",
+       "      <td>0.054314</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.099217</td>\n",
-       "      <td>0.013713</td>\n",
+       "      <td>0.109656</td>\n",
+       "      <td>0.016784</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.099217</td>\n",
-       "      <td>0.013713</td>\n",
+       "      <td>0.109656</td>\n",
+       "      <td>0.016784</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.084680</td>\n",
-       "      <td>0.029714</td>\n",
+       "      <td>0.094767</td>\n",
+       "      <td>0.043454</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.052987</td>\n",
-       "      <td>0.090258</td>\n",
+       "      <td>0.061651</td>\n",
+       "      <td>0.079540</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.105741</td>\n",
-       "      <td>0.103289</td>\n",
+       "      <td>0.113465</td>\n",
+       "      <td>0.110246</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.172701</td>\n",
-       "      <td>0.155497</td>\n",
+       "      <td>0.267356</td>\n",
+       "      <td>0.137744</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1527,14 +1525,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.177476  0.143325\n",
-       "Predictive Parity                                0.010084  0.070287\n",
-       "Equal Opportunity                                0.099217  0.013713\n",
-       "Average Group Difference in False Negative Rate  0.099217  0.013713\n",
-       "Equalized Odds                                   0.084680  0.029714\n",
-       "Conditional Use Accuracy                         0.052987  0.090258\n",
-       "Average Group Difference in Accuracy             0.105741  0.103289\n",
-       "Treatment Equality                               0.172701  0.155497"
+       "Statistical Parity                               0.184497  0.167118\n",
+       "Predictive Parity                                0.025649  0.054314\n",
+       "Equal Opportunity                                0.109656  0.016784\n",
+       "Average Group Difference in False Negative Rate  0.109656  0.016784\n",
+       "Equalized Odds                                   0.094767  0.043454\n",
+       "Conditional Use Accuracy                         0.061651  0.079540\n",
+       "Average Group Difference in Accuracy             0.113465  0.110246\n",
+       "Treatment Equality                               0.267356  0.137744"
       ]
      },
      "execution_count": 13,
@@ -1586,38 +1584,38 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.860360</td>\n",
-       "      <td>0.860934</td>\n",
+       "      <td>0.854464</td>\n",
+       "      <td>0.854791</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.779087</td>\n",
-       "      <td>0.773247</td>\n",
+       "      <td>0.771340</td>\n",
+       "      <td>0.783518</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.681130</td>\n",
-       "      <td>0.675583</td>\n",
+       "      <td>0.668037</td>\n",
+       "      <td>0.680713</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.596944</td>\n",
-       "      <td>0.595141</td>\n",
+       "      <td>0.579837</td>\n",
+       "      <td>0.588426</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.750928</td>\n",
-       "      <td>0.764706</td>\n",
+       "      <td>0.735500</td>\n",
+       "      <td>0.718358</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.623203</td>\n",
-       "      <td>0.605065</td>\n",
+       "      <td>0.611910</td>\n",
+       "      <td>0.646817</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.908998</td>\n",
-       "      <td>0.895994</td>\n",
+       "      <td>0.903369</td>\n",
+       "      <td>0.889127</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1625,13 +1623,13 @@
       ],
       "text/plain": [
        "                   original   updated\n",
-       "Accuracy           0.860360  0.860934\n",
-       "Balanced Accuracy  0.779087  0.773247\n",
-       "F1 score           0.681130  0.675583\n",
-       "MCC                0.596944  0.595141\n",
-       "Precision          0.750928  0.764706\n",
-       "Recall             0.623203  0.605065\n",
-       "ROC AUC            0.908998  0.895994"
+       "Accuracy           0.854464  0.854791\n",
+       "Balanced Accuracy  0.771340  0.783518\n",
+       "F1 score           0.668037  0.680713\n",
+       "MCC                0.579837  0.588426\n",
+       "Precision          0.735500  0.718358\n",
+       "Recall             0.611910  0.646817\n",
+       "ROC AUC            0.903369  0.889127"
       ]
      },
      "execution_count": 14,
@@ -1683,38 +1681,38 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.857178</td>\n",
-       "      <td>0.858570</td>\n",
+       "      <td>0.852756</td>\n",
+       "      <td>0.850708</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.773707</td>\n",
-       "      <td>0.767584</td>\n",
+       "      <td>0.767633</td>\n",
+       "      <td>0.776492</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.672795</td>\n",
-       "      <td>0.667437</td>\n",
+       "      <td>0.662664</td>\n",
+       "      <td>0.670284</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.586992</td>\n",
-       "      <td>0.586832</td>\n",
+       "      <td>0.573967</td>\n",
+       "      <td>0.575758</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.744601</td>\n",
-       "      <td>0.763100</td>\n",
+       "      <td>0.733389</td>\n",
+       "      <td>0.710779</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.613621</td>\n",
-       "      <td>0.593087</td>\n",
+       "      <td>0.604381</td>\n",
+       "      <td>0.634155</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.905340</td>\n",
-       "      <td>0.891552</td>\n",
+       "      <td>0.900784</td>\n",
+       "      <td>0.884691</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1722,13 +1720,13 @@
       ],
       "text/plain": [
        "                   original   updated\n",
-       "Accuracy           0.857178  0.858570\n",
-       "Balanced Accuracy  0.773707  0.767584\n",
-       "F1 score           0.672795  0.667437\n",
-       "MCC                0.586992  0.586832\n",
-       "Precision          0.744601  0.763100\n",
-       "Recall             0.613621  0.593087\n",
-       "ROC AUC            0.905340  0.891552"
+       "Accuracy           0.852756  0.850708\n",
+       "Balanced Accuracy  0.767633  0.776492\n",
+       "F1 score           0.662664  0.670284\n",
+       "MCC                0.573967  0.575758\n",
+       "Precision          0.733389  0.710779\n",
+       "Recall             0.604381  0.634155\n",
+       "ROC AUC            0.900784  0.884691"
       ]
      },
      "execution_count": 15,
@@ -2187,16 +2185,13 @@
        "  /* fitted */\n",
        "  background-color: var(--sklearn-color-fitted-level-3);\n",
        "}\n",
-       "</style><div id=\"sk-container-id-2\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x30d33eb30&gt;,\n",
-       "                      estimator=RandomForestClassifier(),\n",
-       "                      nu=3.5460871351809455e-05)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-4\" type=\"checkbox\" ><label for=\"sk-estimator-id-4\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;ExponentiatedGradient<span class=\"sk-estimator-doc-link fitted\">i<span>Fitted</span></span></label><div class=\"sk-toggleable__content fitted\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x30d33eb30&gt;,\n",
-       "                      estimator=RandomForestClassifier(),\n",
-       "                      nu=3.5460871351809455e-05)</pre></div> </div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-5\" type=\"checkbox\" ><label for=\"sk-estimator-id-5\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">estimator: RandomForestClassifier</label><div class=\"sk-toggleable__content fitted\"><pre>RandomForestClassifier()</pre></div> </div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-6\" type=\"checkbox\" ><label for=\"sk-estimator-id-6\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;RandomForestClassifier<a class=\"sk-estimator-doc-link fitted\" rel=\"noreferrer\" target=\"_blank\" href=\"https://scikit-learn.org/1.4/modules/generated/sklearn.ensemble.RandomForestClassifier.html\">?<span>Documentation for RandomForestClassifier</span></a></label><div class=\"sk-toggleable__content fitted\"><pre>RandomForestClassifier()</pre></div> </div></div></div></div></div></div></div></div></div>"
+       "</style><div id=\"sk-container-id-2\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x164916e30&gt;,\n",
+       "                      estimator=RandomForestClassifier(), nu=0.0)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-4\" type=\"checkbox\" ><label for=\"sk-estimator-id-4\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;ExponentiatedGradient<span class=\"sk-estimator-doc-link fitted\">i<span>Fitted</span></span></label><div class=\"sk-toggleable__content fitted\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x164916e30&gt;,\n",
+       "                      estimator=RandomForestClassifier(), nu=0.0)</pre></div> </div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-5\" type=\"checkbox\" ><label for=\"sk-estimator-id-5\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">estimator: RandomForestClassifier</label><div class=\"sk-toggleable__content fitted\"><pre>RandomForestClassifier()</pre></div> </div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-6\" type=\"checkbox\" ><label for=\"sk-estimator-id-6\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;RandomForestClassifier<a class=\"sk-estimator-doc-link fitted\" rel=\"noreferrer\" target=\"_blank\" href=\"https://scikit-learn.org/1.4/modules/generated/sklearn.ensemble.RandomForestClassifier.html\">?<span>Documentation for RandomForestClassifier</span></a></label><div class=\"sk-toggleable__content fitted\"><pre>RandomForestClassifier()</pre></div> </div></div></div></div></div></div></div></div></div>"
       ],
       "text/plain": [
-       "ExponentiatedGradient(constraints=<fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x30d33eb30>,\n",
-       "                      estimator=RandomForestClassifier(),\n",
-       "                      nu=3.5460871351809455e-05)"
+       "ExponentiatedGradient(constraints=<fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x164916e30>,\n",
+       "                      estimator=RandomForestClassifier(), nu=0.0)"
       ]
      },
      "execution_count": 16,
@@ -2205,8 +2200,8 @@
     }
    ],
    "source": [
-    "mitagator = ExponentiatedGradient(RandomForestClassifier(),TruePositiveRateParity())\n",
-    "mitagator.fit(X=train['data'],y=train['target'],sensitive_features=train['data']['sex'])"
+    "mitagator = ExponentiatedGradient(RandomForestClassifier(), TruePositiveRateParity())\n",
+    "mitagator.fit(X=train['data'], y=train['target'], sensitive_features=train['data']['sex'])"
    ]
   },
   {
@@ -2249,78 +2244,78 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.999918</td>\n",
-       "      <td>0.855540</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.853820</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.999888</td>\n",
-       "      <td>0.770050</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.769271</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.667546</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.665292</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.999775</td>\n",
-       "      <td>0.581346</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.577209</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.742869</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.735794</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.999829</td>\n",
-       "      <td>0.606092</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.607118</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.999888</td>\n",
-       "      <td>0.770050</td>\n",
+       "      <td>1.000000</td>\n",
+       "      <td>0.769271</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.194762</td>\n",
-       "      <td>0.180079</td>\n",
+       "      <td>0.194578</td>\n",
+       "      <td>0.186343</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.000202</td>\n",
-       "      <td>0.011680</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.022945</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.001131</td>\n",
-       "      <td>0.122334</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.120878</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.001131</td>\n",
-       "      <td>0.122334</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.120878</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.000610</td>\n",
-       "      <td>0.096742</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.100165</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.000170</td>\n",
-       "      <td>0.053005</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.059111</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.000062</td>\n",
-       "      <td>0.104866</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.111134</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>1.000000</td>\n",
-       "      <td>0.203247</td>\n",
+       "      <td>0.000000</td>\n",
+       "      <td>0.279414</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -2328,21 +2323,21 @@
       ],
       "text/plain": [
        "                                                    train      test\n",
-       "Accuracy                                         0.999918  0.855540\n",
-       "Balanced Accuracy                                0.999888  0.770050\n",
-       "F1 score                                         0.999829  0.667546\n",
-       "MCC                                              0.999775  0.581346\n",
-       "Precision                                        0.999829  0.742869\n",
-       "Recall                                           0.999829  0.606092\n",
-       "ROC AUC                                          0.999888  0.770050\n",
-       "Statistical Parity                               0.194762  0.180079\n",
-       "Predictive Parity                                0.000202  0.011680\n",
-       "Equal Opportunity                                0.001131  0.122334\n",
-       "Average Group Difference in False Negative Rate  0.001131  0.122334\n",
-       "Equalized Odds                                   0.000610  0.096742\n",
-       "Conditional Use Accuracy                         0.000170  0.053005\n",
-       "Average Group Difference in Accuracy             0.000062  0.104866\n",
-       "Treatment Equality                               1.000000  0.203247"
+       "Accuracy                                         1.000000  0.853820\n",
+       "Balanced Accuracy                                1.000000  0.769271\n",
+       "F1 score                                         1.000000  0.665292\n",
+       "MCC                                              1.000000  0.577209\n",
+       "Precision                                        1.000000  0.735794\n",
+       "Recall                                           1.000000  0.607118\n",
+       "ROC AUC                                          1.000000  0.769271\n",
+       "Statistical Parity                               0.194578  0.186343\n",
+       "Predictive Parity                                0.000000  0.022945\n",
+       "Equal Opportunity                                0.000000  0.120878\n",
+       "Average Group Difference in False Negative Rate  0.000000  0.120878\n",
+       "Equalized Odds                                   0.000000  0.100165\n",
+       "Conditional Use Accuracy                         0.000000  0.059111\n",
+       "Average Group Difference in Accuracy             0.000000  0.111134\n",
+       "Treatment Equality                               0.000000  0.279414"
       ]
      },
      "execution_count": 17,
diff --git a/examples/high-dim_fairlearn_comparision.ipynb b/examples/high-dim_fairlearn_comparision.ipynb
index 23d48ae..95ca804 100644
--- a/examples/high-dim_fairlearn_comparision.ipynb
+++ b/examples/high-dim_fairlearn_comparision.ipynb
@@ -58,13 +58,13 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "/Users/chris.russell/code/oxon-fair/src/oxonfair/utils/dataset_loader.py:202: SettingWithCopyWarning: \n",
+      "/Users/chris.russell/code/oxon-fair/src/oxonfair/utils/dataset_loader.py:216: SettingWithCopyWarning: \n",
       "A value is trying to be set on a copy of a slice from a DataFrame.\n",
       "Try using .loc[row_indexer,col_indexer] = value instead\n",
       "\n",
       "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
       "  X[X.isnull()] = -1\n",
-      "/Users/chris.russell/code/oxon-fair/src/oxonfair/utils/dataset_loader.py:202: SettingWithCopyWarning: \n",
+      "/Users/chris.russell/code/oxon-fair/src/oxonfair/utils/dataset_loader.py:216: SettingWithCopyWarning: \n",
       "A value is trying to be set on a copy of a slice from a DataFrame\n",
       "\n",
       "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
@@ -171,43 +171,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.016250</td>\n",
-       "      <td>0.014167</td>\n",
+       "      <td>0.025708</td>\n",
+       "      <td>0.084434</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.005435</td>\n",
-       "      <td>0.084615</td>\n",
+       "      <td>0.031401</td>\n",
+       "      <td>0.095238</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.064935</td>\n",
-       "      <td>0.011255</td>\n",
+       "      <td>0.150649</td>\n",
+       "      <td>0.009524</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.064935</td>\n",
-       "      <td>0.011255</td>\n",
+       "      <td>0.150649</td>\n",
+       "      <td>0.009524</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.033288</td>\n",
-       "      <td>0.016807</td>\n",
+       "      <td>0.075635</td>\n",
+       "      <td>0.020452</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.009843</td>\n",
-       "      <td>0.044293</td>\n",
+       "      <td>0.054287</td>\n",
+       "      <td>0.070252</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.012500</td>\n",
-       "      <td>0.019583</td>\n",
+       "      <td>0.065920</td>\n",
+       "      <td>0.057193</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.050000</td>\n",
-       "      <td>0.245455</td>\n",
+       "      <td>0.111111</td>\n",
+       "      <td>0.234848</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -215,14 +215,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.016250  0.014167\n",
-       "Predictive Parity                                0.005435  0.084615\n",
-       "Equal Opportunity                                0.064935  0.011255\n",
-       "Average Group Difference in False Negative Rate  0.064935  0.011255\n",
-       "Equalized Odds                                   0.033288  0.016807\n",
-       "Conditional Use Accuracy                         0.009843  0.044293\n",
-       "Average Group Difference in Accuracy             0.012500  0.019583\n",
-       "Treatment Equality                               0.050000  0.245455"
+       "Statistical Parity                               0.025708  0.084434\n",
+       "Predictive Parity                                0.031401  0.095238\n",
+       "Equal Opportunity                                0.150649  0.009524\n",
+       "Average Group Difference in False Negative Rate  0.150649  0.009524\n",
+       "Equalized Odds                                   0.075635  0.020452\n",
+       "Conditional Use Accuracy                         0.054287  0.070252\n",
+       "Average Group Difference in Accuracy             0.065920  0.057193\n",
+       "Treatment Equality                               0.111111  0.234848"
       ]
      },
      "execution_count": 5,
@@ -281,43 +281,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.059044</td>\n",
-       "      <td>0.064624</td>\n",
+       "      <td>0.041601</td>\n",
+       "      <td>0.101147</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.163399</td>\n",
-       "      <td>0.108065</td>\n",
+       "      <td>0.185464</td>\n",
+       "      <td>0.065831</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.147059</td>\n",
-       "      <td>0.176471</td>\n",
+       "      <td>0.058824</td>\n",
+       "      <td>0.205882</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.147059</td>\n",
-       "      <td>0.176471</td>\n",
+       "      <td>0.058824</td>\n",
+       "      <td>0.205882</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.093486</td>\n",
-       "      <td>0.107781</td>\n",
+       "      <td>0.036109</td>\n",
+       "      <td>0.114259</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.098748</td>\n",
-       "      <td>0.075119</td>\n",
+       "      <td>0.113700</td>\n",
+       "      <td>0.038816</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.004747</td>\n",
-       "      <td>0.011326</td>\n",
+       "      <td>0.021453</td>\n",
+       "      <td>0.017707</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.405983</td>\n",
-       "      <td>0.550802</td>\n",
+       "      <td>0.215278</td>\n",
+       "      <td>0.212121</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -325,14 +325,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.059044  0.064624\n",
-       "Predictive Parity                                0.163399  0.108065\n",
-       "Equal Opportunity                                0.147059  0.176471\n",
-       "Average Group Difference in False Negative Rate  0.147059  0.176471\n",
-       "Equalized Odds                                   0.093486  0.107781\n",
-       "Conditional Use Accuracy                         0.098748  0.075119\n",
-       "Average Group Difference in Accuracy             0.004747  0.011326\n",
-       "Treatment Equality                               0.405983  0.550802"
+       "Statistical Parity                               0.041601  0.101147\n",
+       "Predictive Parity                                0.185464  0.065831\n",
+       "Equal Opportunity                                0.058824  0.205882\n",
+       "Average Group Difference in False Negative Rate  0.058824  0.205882\n",
+       "Equalized Odds                                   0.036109  0.114259\n",
+       "Conditional Use Accuracy                         0.113700  0.038816\n",
+       "Average Group Difference in Accuracy             0.021453  0.017707\n",
+       "Treatment Equality                               0.215278  0.212121"
       ]
      },
      "execution_count": 6,
@@ -391,38 +391,38 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.893548</td>\n",
-       "      <td>0.903226</td>\n",
+       "      <td>0.922353</td>\n",
+       "      <td>0.929412</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.789074</td>\n",
-       "      <td>0.826993</td>\n",
+       "      <td>0.775210</td>\n",
+       "      <td>0.821078</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.713043</td>\n",
-       "      <td>0.758065</td>\n",
+       "      <td>0.697248</td>\n",
+       "      <td>0.750000</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.667074</td>\n",
-       "      <td>0.703429</td>\n",
+       "      <td>0.683479</td>\n",
+       "      <td>0.718412</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.872340</td>\n",
-       "      <td>0.839286</td>\n",
+       "      <td>0.926829</td>\n",
+       "      <td>0.865385</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.602941</td>\n",
-       "      <td>0.691176</td>\n",
+       "      <td>0.558824</td>\n",
+       "      <td>0.661765</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.897302</td>\n",
-       "      <td>0.881745</td>\n",
+       "      <td>0.933844</td>\n",
+       "      <td>0.912300</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -430,13 +430,13 @@
       ],
       "text/plain": [
        "                   original   updated\n",
-       "Accuracy           0.893548  0.903226\n",
-       "Balanced Accuracy  0.789074  0.826993\n",
-       "F1 score           0.713043  0.758065\n",
-       "MCC                0.667074  0.703429\n",
-       "Precision          0.872340  0.839286\n",
-       "Recall             0.602941  0.691176\n",
-       "ROC AUC            0.897302  0.881745"
+       "Accuracy           0.922353  0.929412\n",
+       "Balanced Accuracy  0.775210  0.821078\n",
+       "F1 score           0.697248  0.750000\n",
+       "MCC                0.683479  0.718412\n",
+       "Precision          0.926829  0.865385\n",
+       "Recall             0.558824  0.661765\n",
+       "ROC AUC            0.933844  0.912300"
       ]
      },
      "execution_count": 7,
@@ -495,38 +495,38 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.877419</td>\n",
-       "      <td>0.874194</td>\n",
+       "      <td>0.905882</td>\n",
+       "      <td>0.903529</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.757596</td>\n",
-       "      <td>0.771390</td>\n",
+       "      <td>0.741597</td>\n",
+       "      <td>0.769958</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.660714</td>\n",
-       "      <td>0.672269</td>\n",
+       "      <td>0.629630</td>\n",
+       "      <td>0.655462</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.610892</td>\n",
-       "      <td>0.605831</td>\n",
+       "      <td>0.606665</td>\n",
+       "      <td>0.609107</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.840909</td>\n",
-       "      <td>0.784314</td>\n",
+       "      <td>0.850000</td>\n",
+       "      <td>0.764706</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.544118</td>\n",
-       "      <td>0.588235</td>\n",
+       "      <td>0.500000</td>\n",
+       "      <td>0.573529</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.898396</td>\n",
-       "      <td>0.853610</td>\n",
+       "      <td>0.900972</td>\n",
+       "      <td>0.862539</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -534,13 +534,13 @@
       ],
       "text/plain": [
        "                   original   updated\n",
-       "Accuracy           0.877419  0.874194\n",
-       "Balanced Accuracy  0.757596  0.771390\n",
-       "F1 score           0.660714  0.672269\n",
-       "MCC                0.610892  0.605831\n",
-       "Precision          0.840909  0.784314\n",
-       "Recall             0.544118  0.588235\n",
-       "ROC AUC            0.898396  0.853610"
+       "Accuracy           0.905882  0.903529\n",
+       "Balanced Accuracy  0.741597  0.769958\n",
+       "F1 score           0.629630  0.655462\n",
+       "MCC                0.606665  0.609107\n",
+       "Precision          0.850000  0.764706\n",
+       "Recall             0.500000  0.573529\n",
+       "ROC AUC            0.900972  0.862539"
       ]
      },
      "execution_count": 8,
@@ -1006,7 +1006,7 @@
        "  /* fitted */\n",
        "  background-color: var(--sklearn-color-fitted-level-3);\n",
        "}\n",
-       "</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x3094d83a0&gt;,\n",
+       "</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x1659d0130&gt;,\n",
        "                      estimator=XGBClassifier(base_score=None, booster=None,\n",
        "                                              callbacks=None,\n",
        "                                              colsample_bylevel=None,\n",
@@ -1032,7 +1032,7 @@
        "                                              n_estimators=None, n_jobs=None,\n",
        "                                              num_parallel_tree=None,\n",
        "                                              random_state=None, ...),\n",
-       "                      nu=0.0)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" ><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;ExponentiatedGradient<span class=\"sk-estimator-doc-link fitted\">i<span>Fitted</span></span></label><div class=\"sk-toggleable__content fitted\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x3094d83a0&gt;,\n",
+       "                      nu=0.0)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" ><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;ExponentiatedGradient<span class=\"sk-estimator-doc-link fitted\">i<span>Fitted</span></span></label><div class=\"sk-toggleable__content fitted\"><pre>ExponentiatedGradient(constraints=&lt;fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x1659d0130&gt;,\n",
        "                      estimator=XGBClassifier(base_score=None, booster=None,\n",
        "                                              callbacks=None,\n",
        "                                              colsample_bylevel=None,\n",
@@ -1081,7 +1081,7 @@
        "              num_parallel_tree=None, random_state=None, ...)</pre></div> </div></div></div></div></div></div></div></div></div>"
       ],
       "text/plain": [
-       "ExponentiatedGradient(constraints=<fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x3094d83a0>,\n",
+       "ExponentiatedGradient(constraints=<fairlearn.reductions._moments.utility_parity.TruePositiveRateParity object at 0x1659d0130>,\n",
        "                      estimator=XGBClassifier(base_score=None, booster=None,\n",
        "                                              callbacks=None,\n",
        "                                              colsample_bylevel=None,\n",
@@ -1169,77 +1169,77 @@
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.877419</td>\n",
+       "      <td>0.905882</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.757596</td>\n",
+       "      <td>0.741597</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.660714</td>\n",
+       "      <td>0.629630</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.610892</td>\n",
+       "      <td>0.606665</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.840909</td>\n",
+       "      <td>0.850000</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.544118</td>\n",
+       "      <td>0.500000</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.757596</td>\n",
+       "      <td>0.741597</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.006611</td>\n",
-       "      <td>0.059044</td>\n",
+       "      <td>0.088807</td>\n",
+       "      <td>0.041601</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.163399</td>\n",
+       "      <td>0.185464</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.147059</td>\n",
+       "      <td>0.058824</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.147059</td>\n",
+       "      <td>0.058824</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.093486</td>\n",
+       "      <td>0.036109</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.098748</td>\n",
+       "      <td>0.113700</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.004747</td>\n",
+       "      <td>0.021453</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.405983</td>\n",
+       "      <td>0.215278</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1247,21 +1247,21 @@
       ],
       "text/plain": [
        "                                                    train      test\n",
-       "Accuracy                                         1.000000  0.877419\n",
-       "Balanced Accuracy                                1.000000  0.757596\n",
-       "F1 score                                         1.000000  0.660714\n",
-       "MCC                                              1.000000  0.610892\n",
-       "Precision                                        1.000000  0.840909\n",
-       "Recall                                           1.000000  0.544118\n",
-       "ROC AUC                                          1.000000  0.757596\n",
-       "Statistical Parity                               0.006611  0.059044\n",
-       "Predictive Parity                                0.000000  0.163399\n",
-       "Equal Opportunity                                0.000000  0.147059\n",
-       "Average Group Difference in False Negative Rate  0.000000  0.147059\n",
-       "Equalized Odds                                   0.000000  0.093486\n",
-       "Conditional Use Accuracy                         0.000000  0.098748\n",
-       "Average Group Difference in Accuracy             0.000000  0.004747\n",
-       "Treatment Equality                               0.000000  0.405983"
+       "Accuracy                                         1.000000  0.905882\n",
+       "Balanced Accuracy                                1.000000  0.741597\n",
+       "F1 score                                         1.000000  0.629630\n",
+       "MCC                                              1.000000  0.606665\n",
+       "Precision                                        1.000000  0.850000\n",
+       "Recall                                           1.000000  0.500000\n",
+       "ROC AUC                                          1.000000  0.741597\n",
+       "Statistical Parity                               0.088807  0.041601\n",
+       "Predictive Parity                                0.000000  0.185464\n",
+       "Equal Opportunity                                0.000000  0.058824\n",
+       "Average Group Difference in False Negative Rate  0.000000  0.058824\n",
+       "Equalized Odds                                   0.000000  0.036109\n",
+       "Conditional Use Accuracy                         0.000000  0.113700\n",
+       "Average Group Difference in Accuracy             0.000000  0.021453\n",
+       "Treatment Equality                               0.000000  0.215278"
       ]
      },
      "execution_count": 10,
@@ -1272,7 +1272,8 @@
    "source": [
     "def eval(train, classifier=mitagator):\n",
     "    return pd.concat((performance.evaluate(train['target'], classifier.predict(train['data'])),\n",
-    "                      performance.evaluate_fairness(train['target'], classifier.predict(train['data']), train['groups'])),axis=0)\n",
+    "                      performance.evaluate_fairness(train['target'], classifier.predict(train['data']),\n",
+    "                                                    train['data'][train['groups']])),axis=0)\n",
     "\n",
     "out = pd.concat((eval(train), eval(test)), axis=1)\n",
     "out.columns = ['train', 'test']\n",
@@ -1327,77 +1328,77 @@
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.877419</td>\n",
+       "      <td>0.905882</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.757596</td>\n",
+       "      <td>0.741597</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.660714</td>\n",
+       "      <td>0.629630</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.610892</td>\n",
+       "      <td>0.606665</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.840909</td>\n",
+       "      <td>0.850000</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.544118</td>\n",
+       "      <td>0.500000</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
        "      <td>1.000000</td>\n",
-       "      <td>0.757596</td>\n",
+       "      <td>0.741597</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.006611</td>\n",
-       "      <td>0.059044</td>\n",
+       "      <td>0.088807</td>\n",
+       "      <td>0.041601</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.163399</td>\n",
+       "      <td>0.185464</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.147059</td>\n",
+       "      <td>0.058824</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.147059</td>\n",
+       "      <td>0.058824</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.093486</td>\n",
+       "      <td>0.036109</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.098748</td>\n",
+       "      <td>0.113700</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.004747</td>\n",
+       "      <td>0.021453</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
        "      <td>0.000000</td>\n",
-       "      <td>0.405983</td>\n",
+       "      <td>0.215278</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1405,21 +1406,21 @@
       ],
       "text/plain": [
        "                                                    train      test\n",
-       "Accuracy                                         1.000000  0.877419\n",
-       "Balanced Accuracy                                1.000000  0.757596\n",
-       "F1 score                                         1.000000  0.660714\n",
-       "MCC                                              1.000000  0.610892\n",
-       "Precision                                        1.000000  0.840909\n",
-       "Recall                                           1.000000  0.544118\n",
-       "ROC AUC                                          1.000000  0.757596\n",
-       "Statistical Parity                               0.006611  0.059044\n",
-       "Predictive Parity                                0.000000  0.163399\n",
-       "Equal Opportunity                                0.000000  0.147059\n",
-       "Average Group Difference in False Negative Rate  0.000000  0.147059\n",
-       "Equalized Odds                                   0.000000  0.093486\n",
-       "Conditional Use Accuracy                         0.000000  0.098748\n",
-       "Average Group Difference in Accuracy             0.000000  0.004747\n",
-       "Treatment Equality                               0.000000  0.405983"
+       "Accuracy                                         1.000000  0.905882\n",
+       "Balanced Accuracy                                1.000000  0.741597\n",
+       "F1 score                                         1.000000  0.629630\n",
+       "MCC                                              1.000000  0.606665\n",
+       "Precision                                        1.000000  0.850000\n",
+       "Recall                                           1.000000  0.500000\n",
+       "ROC AUC                                          1.000000  0.741597\n",
+       "Statistical Parity                               0.088807  0.041601\n",
+       "Predictive Parity                                0.000000  0.185464\n",
+       "Equal Opportunity                                0.000000  0.058824\n",
+       "Average Group Difference in False Negative Rate  0.000000  0.058824\n",
+       "Equalized Odds                                   0.000000  0.036109\n",
+       "Conditional Use Accuracy                         0.000000  0.113700\n",
+       "Average Group Difference in Accuracy             0.000000  0.021453\n",
+       "Treatment Equality                               0.000000  0.215278"
       ]
      },
      "execution_count": 11,
diff --git a/examples/multi_group_fairlearn_comparision.ipynb b/examples/multi_group_fairlearn_comparision.ipynb
index ed5d4aa..e9954e9 100644
--- a/examples/multi_group_fairlearn_comparision.ipynb
+++ b/examples/multi_group_fairlearn_comparision.ipynb
@@ -23,20 +23,12 @@
     }
    },
    "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
-      "  from .autonotebook import tqdm as notebook_tqdm\n"
-     ]
-    },
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
       "Training time of xgboost without fairness\n",
-      "0.635854082996957\n"
+      "0.681041999999934\n"
      ]
     }
    ],
@@ -52,7 +44,9 @@
     "import fairlearn\n",
     "import time\n",
     "import pandas as pd\n",
-    "train_data, val_data, test_data = dataset_loader.adult('race',train_proportion=0.7,test_proportion=0.3)\n",
+    "train_data, val_data, test_data = dataset_loader.adult('race',train_proportion=0.7,test_proportion=0.3, replace_groups={'a':'a'})\n",
+    "\n",
+    "#Replace groups here is a null op that ensures that groups is not a string, and we can modify it later.\n",
     "\n",
     "start=time.perf_counter()\n",
     "predictor = xgboost.XGBClassifier().fit(X=train_data['data'], y=train_data['target'])\n",
@@ -86,7 +80,7 @@
     },
     {
      "data": {
-      "image/png": "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",
+      "image/png": "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",
       "text/plain": [
        "<Figure size 640x480 with 1 Axes>"
       ]
@@ -130,252 +124,7 @@
      "shell.execute_reply": "2024-06-17T19:21:25.414759Z"
     }
    },
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n"
-     ]
-    },
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:214: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.pos_basis[i][\"+\", e, g] = 1\n",
-      "/opt/miniconda3/envs/ag/lib/python3.10/site-packages/fairlearn/reductions/_moments/utility_parity.py:215: FutureWarning: ChainedAssignmentError: behaviour will change in pandas 3.0!\n",
-      "You are setting values through chained assignment. Currently this works in certain cases, but when using Copy-on-Write (which will become the default behaviour in pandas 3.0) this will never work to update the original DataFrame or Series, because the intermediate object on which we are setting values will behave as a copy.\n",
-      "A typical example is when you are setting values in a column of a DataFrame, like:\n",
-      "\n",
-      "df[\"col\"][row_indexer] = value\n",
-      "\n",
-      "Use `df.loc[row_indexer, \"col\"] = values` instead, to perform the assignment in a single step and ensure this keeps updating the original `df`.\n",
-      "\n",
-      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
-      "\n",
-      "  self.neg_basis[i][\"-\", e, g] = 1\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "for i,name in enumerate ((' Amer-Indian-Eskimo', ' Asian-Pac-Islander', ' Black', ' Other')):\n",
     "    # This looks like a bug, why are we merging ' Other' with itself.\n",
@@ -463,39 +212,39 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
+       "      <td>0.871826</td>\n",
+       "      <td>0.020394</td>\n",
+       "      <td>43.286311</td>\n",
        "      <td>0.868482</td>\n",
-       "      <td>0.045112</td>\n",
-       "      <td>95.299039</td>\n",
-       "      <td>0.868687</td>\n",
-       "      <td>0.046261</td>\n",
-       "      <td>16.292495</td>\n",
+       "      <td>0.020783</td>\n",
+       "      <td>44.239217</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
-       "      <td>0.868073</td>\n",
-       "      <td>0.029650</td>\n",
-       "      <td>1.049697</td>\n",
-       "      <td>0.870461</td>\n",
-       "      <td>0.015604</td>\n",
-       "      <td>10.474953</td>\n",
+       "      <td>0.871622</td>\n",
+       "      <td>0.017975</td>\n",
+       "      <td>0.754434</td>\n",
+       "      <td>0.871826</td>\n",
+       "      <td>0.019802</td>\n",
+       "      <td>30.235781</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
-       "      <td>0.868209</td>\n",
-       "      <td>0.009898</td>\n",
-       "      <td>0.043179</td>\n",
-       "      <td>0.871280</td>\n",
-       "      <td>0.009638</td>\n",
-       "      <td>3.923120</td>\n",
+       "      <td>0.871417</td>\n",
+       "      <td>0.022778</td>\n",
+       "      <td>0.088691</td>\n",
+       "      <td>0.871417</td>\n",
+       "      <td>0.013146</td>\n",
+       "      <td>25.762889</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
-       "      <td>0.870461</td>\n",
-       "      <td>0.013591</td>\n",
-       "      <td>0.026398</td>\n",
-       "      <td>0.870598</td>\n",
-       "      <td>0.027901</td>\n",
-       "      <td>4.758493</td>\n",
+       "      <td>0.871622</td>\n",
+       "      <td>0.006771</td>\n",
+       "      <td>0.051017</td>\n",
+       "      <td>0.871485</td>\n",
+       "      <td>0.010087</td>\n",
+       "      <td>19.135028</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -503,16 +252,16 @@
       ],
       "text/plain": [
        "   Accuracy  Demographic Parity       Time  Accuracy  Demographic Parity  \\\n",
-       "0  0.868482            0.045112  95.299039  0.868687            0.046261   \n",
-       "1  0.868073            0.029650   1.049697  0.870461            0.015604   \n",
-       "2  0.868209            0.009898   0.043179  0.871280            0.009638   \n",
-       "3  0.870461            0.013591   0.026398  0.870598            0.027901   \n",
+       "0  0.871826            0.020394  43.286311  0.868482            0.020783   \n",
+       "1  0.871622            0.017975   0.754434  0.871826            0.019802   \n",
+       "2  0.871417            0.022778   0.088691  0.871417            0.013146   \n",
+       "3  0.871622            0.006771   0.051017  0.871485            0.010087   \n",
        "\n",
        "        Time  \n",
-       "0  16.292495  \n",
-       "1  10.474953  \n",
-       "2   3.923120  \n",
-       "3   4.758493  "
+       "0  44.239217  \n",
+       "1  30.235781  \n",
+       "2  25.762889  \n",
+       "3  19.135028  "
       ]
      },
      "execution_count": 6,
diff --git a/examples/quickstart_xgboost.ipynb b/examples/quickstart_xgboost.ipynb
index 726dc0d..a4a3418 100644
--- a/examples/quickstart_xgboost.ipynb
+++ b/examples/quickstart_xgboost.ipynb
@@ -138,7 +138,7 @@
     {
      "data": {
       "text/plain": [
-       "array([1, 0, 1, ..., 0, 0, 0])"
+       "array([0, 0, 0, ..., 0, 0, 0])"
       ]
      },
      "execution_count": 4,
@@ -191,38 +191,38 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.868315</td>\n",
-       "      <td>0.848743</td>\n",
+       "      <td>0.863156</td>\n",
+       "      <td>0.845222</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Balanced Accuracy</th>\n",
-       "      <td>0.797330</td>\n",
-       "      <td>0.768163</td>\n",
+       "      <td>0.789951</td>\n",
+       "      <td>0.759749</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>F1 score</th>\n",
-       "      <td>0.706140</td>\n",
-       "      <td>0.660040</td>\n",
+       "      <td>0.694348</td>\n",
+       "      <td>0.648176</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>MCC</th>\n",
-       "      <td>0.624184</td>\n",
-       "      <td>0.566166</td>\n",
+       "      <td>0.609134</td>\n",
+       "      <td>0.553450</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Precision</th>\n",
-       "      <td>0.757647</td>\n",
-       "      <td>0.714058</td>\n",
+       "      <td>0.745776</td>\n",
+       "      <td>0.710612</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Recall</th>\n",
-       "      <td>0.661191</td>\n",
-       "      <td>0.613621</td>\n",
+       "      <td>0.649555</td>\n",
+       "      <td>0.595825</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>ROC AUC</th>\n",
-       "      <td>0.925353</td>\n",
-       "      <td>0.821574</td>\n",
+       "      <td>0.920422</td>\n",
+       "      <td>0.813340</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -230,13 +230,13 @@
       ],
       "text/plain": [
        "                   original   updated\n",
-       "Accuracy           0.868315  0.848743\n",
-       "Balanced Accuracy  0.797330  0.768163\n",
-       "F1 score           0.706140  0.660040\n",
-       "MCC                0.624184  0.566166\n",
-       "Precision          0.757647  0.714058\n",
-       "Recall             0.661191  0.613621\n",
-       "ROC AUC            0.925353  0.821574"
+       "Accuracy           0.863156  0.845222\n",
+       "Balanced Accuracy  0.789951  0.759749\n",
+       "F1 score           0.694348  0.648176\n",
+       "MCC                0.609134  0.553450\n",
+       "Precision          0.745776  0.710612\n",
+       "Recall             0.649555  0.595825\n",
+       "ROC AUC            0.920422  0.813340"
       ]
      },
      "execution_count": 5,
@@ -289,43 +289,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.192654</td>\n",
-       "      <td>0.013454</td>\n",
+       "      <td>0.194998</td>\n",
+       "      <td>0.008199</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.023061</td>\n",
-       "      <td>0.363074</td>\n",
+       "      <td>0.030115</td>\n",
+       "      <td>0.359417</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.107284</td>\n",
-       "      <td>0.265979</td>\n",
+       "      <td>0.114899</td>\n",
+       "      <td>0.281616</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.107284</td>\n",
-       "      <td>0.265979</td>\n",
+       "      <td>0.114899</td>\n",
+       "      <td>0.281616</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.091696</td>\n",
-       "      <td>0.166141</td>\n",
+       "      <td>0.098056</td>\n",
+       "      <td>0.173844</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.053394</td>\n",
-       "      <td>0.252662</td>\n",
+       "      <td>0.058129</td>\n",
+       "      <td>0.253697</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.100167</td>\n",
-       "      <td>0.042973</td>\n",
+       "      <td>0.104928</td>\n",
+       "      <td>0.048980</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.303553</td>\n",
-       "      <td>5.708978</td>\n",
+       "      <td>0.332490</td>\n",
+       "      <td>5.507195</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -333,14 +333,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.192654  0.013454\n",
-       "Predictive Parity                                0.023061  0.363074\n",
-       "Equal Opportunity                                0.107284  0.265979\n",
-       "Average Group Difference in False Negative Rate  0.107284  0.265979\n",
-       "Equalized Odds                                   0.091696  0.166141\n",
-       "Conditional Use Accuracy                         0.053394  0.252662\n",
-       "Average Group Difference in Accuracy             0.100167  0.042973\n",
-       "Treatment Equality                               0.303553  5.708978"
+       "Statistical Parity                               0.194998  0.008199\n",
+       "Predictive Parity                                0.030115  0.359417\n",
+       "Equal Opportunity                                0.114899  0.281616\n",
+       "Average Group Difference in False Negative Rate  0.114899  0.281616\n",
+       "Equalized Odds                                   0.098056  0.173844\n",
+       "Conditional Use Accuracy                         0.058129  0.253697\n",
+       "Average Group Difference in Accuracy             0.104928  0.048980\n",
+       "Treatment Equality                               0.332490  5.507195"
       ]
      },
      "execution_count": 6,
@@ -419,116 +419,116 @@
        "    <tr>\n",
        "      <th rowspan=\"4\" valign=\"top\">original</th>\n",
        "      <th>Overall</th>\n",
-       "      <td>0.868315</td>\n",
-       "      <td>0.797330</td>\n",
-       "      <td>0.706140</td>\n",
-       "      <td>0.624184</td>\n",
-       "      <td>0.757647</td>\n",
-       "      <td>0.661191</td>\n",
-       "      <td>0.925353</td>\n",
+       "      <td>0.863156</td>\n",
+       "      <td>0.789951</td>\n",
+       "      <td>0.694348</td>\n",
+       "      <td>0.609134</td>\n",
+       "      <td>0.745776</td>\n",
+       "      <td>0.649555</td>\n",
+       "      <td>0.920422</td>\n",
        "      <td>2922.0</td>\n",
        "      <td>9289.0</td>\n",
        "      <td>0.239292</td>\n",
-       "      <td>0.208828</td>\n",
+       "      <td>0.208419</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
-       "      <td>0.935277</td>\n",
-       "      <td>0.775085</td>\n",
-       "      <td>0.657963</td>\n",
-       "      <td>0.632329</td>\n",
-       "      <td>0.777778</td>\n",
-       "      <td>0.570136</td>\n",
-       "      <td>0.941812</td>\n",
+       "      <td>0.933300</td>\n",
+       "      <td>0.766035</td>\n",
+       "      <td>0.643799</td>\n",
+       "      <td>0.618555</td>\n",
+       "      <td>0.772152</td>\n",
+       "      <td>0.552036</td>\n",
+       "      <td>0.938214</td>\n",
        "      <td>442.0</td>\n",
        "      <td>3606.0</td>\n",
        "      <td>0.109190</td>\n",
-       "      <td>0.080040</td>\n",
+       "      <td>0.078063</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
-       "      <td>0.835110</td>\n",
-       "      <td>0.790672</td>\n",
-       "      <td>0.713982</td>\n",
-       "      <td>0.600346</td>\n",
-       "      <td>0.754717</td>\n",
-       "      <td>0.677419</td>\n",
-       "      <td>0.907388</td>\n",
+       "      <td>0.828372</td>\n",
+       "      <td>0.782878</td>\n",
+       "      <td>0.702485</td>\n",
+       "      <td>0.584003</td>\n",
+       "      <td>0.742037</td>\n",
+       "      <td>0.666935</td>\n",
+       "      <td>0.900383</td>\n",
        "      <td>2480.0</td>\n",
        "      <td>5683.0</td>\n",
        "      <td>0.303810</td>\n",
-       "      <td>0.272694</td>\n",
+       "      <td>0.273061</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Maximum difference</th>\n",
-       "      <td>0.100167</td>\n",
-       "      <td>0.015587</td>\n",
-       "      <td>0.056019</td>\n",
-       "      <td>0.031983</td>\n",
-       "      <td>0.023061</td>\n",
-       "      <td>0.107284</td>\n",
-       "      <td>0.034423</td>\n",
+       "      <td>0.104928</td>\n",
+       "      <td>0.016844</td>\n",
+       "      <td>0.058685</td>\n",
+       "      <td>0.034551</td>\n",
+       "      <td>0.030115</td>\n",
+       "      <td>0.114899</td>\n",
+       "      <td>0.037831</td>\n",
        "      <td>2038.0</td>\n",
        "      <td>2077.0</td>\n",
        "      <td>0.194620</td>\n",
-       "      <td>0.192654</td>\n",
+       "      <td>0.194998</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th rowspan=\"4\" valign=\"top\">updated</th>\n",
        "      <th>Overall</th>\n",
-       "      <td>0.848743</td>\n",
-       "      <td>0.768163</td>\n",
-       "      <td>0.660040</td>\n",
-       "      <td>0.566166</td>\n",
-       "      <td>0.714058</td>\n",
-       "      <td>0.613621</td>\n",
-       "      <td>0.821574</td>\n",
+       "      <td>0.845222</td>\n",
+       "      <td>0.759749</td>\n",
+       "      <td>0.648176</td>\n",
+       "      <td>0.553450</td>\n",
+       "      <td>0.710612</td>\n",
+       "      <td>0.595825</td>\n",
+       "      <td>0.813340</td>\n",
        "      <td>2922.0</td>\n",
        "      <td>9289.0</td>\n",
        "      <td>0.239292</td>\n",
-       "      <td>0.205634</td>\n",
+       "      <td>0.200639</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
-       "      <td>0.877470</td>\n",
-       "      <td>0.860754</td>\n",
-       "      <td>0.599354</td>\n",
-       "      <td>0.566152</td>\n",
-       "      <td>0.466080</td>\n",
-       "      <td>0.839367</td>\n",
-       "      <td>0.941812</td>\n",
+       "      <td>0.877964</td>\n",
+       "      <td>0.859046</td>\n",
+       "      <td>0.599026</td>\n",
+       "      <td>0.565086</td>\n",
+       "      <td>0.467089</td>\n",
+       "      <td>0.834842</td>\n",
+       "      <td>0.938214</td>\n",
        "      <td>442.0</td>\n",
        "      <td>3606.0</td>\n",
        "      <td>0.109190</td>\n",
-       "      <td>0.196640</td>\n",
+       "      <td>0.195158</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
-       "      <td>0.834497</td>\n",
-       "      <td>0.760915</td>\n",
-       "      <td>0.677950</td>\n",
-       "      <td>0.589114</td>\n",
-       "      <td>0.829155</td>\n",
-       "      <td>0.573387</td>\n",
-       "      <td>0.907388</td>\n",
+       "      <td>0.828984</td>\n",
+       "      <td>0.751274</td>\n",
+       "      <td>0.662802</td>\n",
+       "      <td>0.574224</td>\n",
+       "      <td>0.826506</td>\n",
+       "      <td>0.553226</td>\n",
+       "      <td>0.900383</td>\n",
        "      <td>2480.0</td>\n",
        "      <td>5683.0</td>\n",
        "      <td>0.303810</td>\n",
-       "      <td>0.210094</td>\n",
+       "      <td>0.203357</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Maximum difference</th>\n",
-       "      <td>0.042973</td>\n",
-       "      <td>0.099839</td>\n",
-       "      <td>0.078596</td>\n",
-       "      <td>0.022962</td>\n",
-       "      <td>0.363074</td>\n",
-       "      <td>0.265979</td>\n",
-       "      <td>0.034423</td>\n",
+       "      <td>0.048980</td>\n",
+       "      <td>0.107772</td>\n",
+       "      <td>0.063776</td>\n",
+       "      <td>0.009138</td>\n",
+       "      <td>0.359417</td>\n",
+       "      <td>0.281616</td>\n",
+       "      <td>0.037831</td>\n",
        "      <td>2038.0</td>\n",
        "      <td>2077.0</td>\n",
        "      <td>0.194620</td>\n",
-       "      <td>0.013454</td>\n",
+       "      <td>0.008199</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -537,25 +537,25 @@
       "text/plain": [
        "                             Accuracy  Balanced Accuracy  F1 score       MCC  \\\n",
        "         Groups                                                                \n",
-       "original Overall             0.868315           0.797330  0.706140  0.624184   \n",
-       "         0                   0.935277           0.775085  0.657963  0.632329   \n",
-       "         1                   0.835110           0.790672  0.713982  0.600346   \n",
-       "         Maximum difference  0.100167           0.015587  0.056019  0.031983   \n",
-       "updated  Overall             0.848743           0.768163  0.660040  0.566166   \n",
-       "         0                   0.877470           0.860754  0.599354  0.566152   \n",
-       "         1                   0.834497           0.760915  0.677950  0.589114   \n",
-       "         Maximum difference  0.042973           0.099839  0.078596  0.022962   \n",
+       "original Overall             0.863156           0.789951  0.694348  0.609134   \n",
+       "         0                   0.933300           0.766035  0.643799  0.618555   \n",
+       "         1                   0.828372           0.782878  0.702485  0.584003   \n",
+       "         Maximum difference  0.104928           0.016844  0.058685  0.034551   \n",
+       "updated  Overall             0.845222           0.759749  0.648176  0.553450   \n",
+       "         0                   0.877964           0.859046  0.599026  0.565086   \n",
+       "         1                   0.828984           0.751274  0.662802  0.574224   \n",
+       "         Maximum difference  0.048980           0.107772  0.063776  0.009138   \n",
        "\n",
        "                             Precision    Recall   ROC AUC  Positive Count  \\\n",
        "         Groups                                                              \n",
-       "original Overall              0.757647  0.661191  0.925353          2922.0   \n",
-       "         0                    0.777778  0.570136  0.941812           442.0   \n",
-       "         1                    0.754717  0.677419  0.907388          2480.0   \n",
-       "         Maximum difference   0.023061  0.107284  0.034423          2038.0   \n",
-       "updated  Overall              0.714058  0.613621  0.821574          2922.0   \n",
-       "         0                    0.466080  0.839367  0.941812           442.0   \n",
-       "         1                    0.829155  0.573387  0.907388          2480.0   \n",
-       "         Maximum difference   0.363074  0.265979  0.034423          2038.0   \n",
+       "original Overall              0.745776  0.649555  0.920422          2922.0   \n",
+       "         0                    0.772152  0.552036  0.938214           442.0   \n",
+       "         1                    0.742037  0.666935  0.900383          2480.0   \n",
+       "         Maximum difference   0.030115  0.114899  0.037831          2038.0   \n",
+       "updated  Overall              0.710612  0.595825  0.813340          2922.0   \n",
+       "         0                    0.467089  0.834842  0.938214           442.0   \n",
+       "         1                    0.826506  0.553226  0.900383          2480.0   \n",
+       "         Maximum difference   0.359417  0.281616  0.037831          2038.0   \n",
        "\n",
        "                             Negative Count  Positive Label Rate  \\\n",
        "         Groups                                                    \n",
@@ -570,14 +570,14 @@
        "\n",
        "                             Positive Prediction Rate  \n",
        "         Groups                                        \n",
-       "original Overall                             0.208828  \n",
-       "         0                                   0.080040  \n",
-       "         1                                   0.272694  \n",
-       "         Maximum difference                  0.192654  \n",
-       "updated  Overall                             0.205634  \n",
-       "         0                                   0.196640  \n",
-       "         1                                   0.210094  \n",
-       "         Maximum difference                  0.013454  "
+       "original Overall                             0.208419  \n",
+       "         0                                   0.078063  \n",
+       "         1                                   0.273061  \n",
+       "         Maximum difference                  0.194998  \n",
+       "updated  Overall                             0.200639  \n",
+       "         0                                   0.195158  \n",
+       "         1                                   0.203357  \n",
+       "         Maximum difference                  0.008199  "
       ]
      },
      "execution_count": 7,
@@ -673,43 +673,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.103547</td>\n",
-       "      <td>0.049493</td>\n",
+       "      <td>0.125058</td>\n",
+       "      <td>0.073183</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.074152</td>\n",
-       "      <td>0.184299</td>\n",
+       "      <td>0.013161</td>\n",
+       "      <td>0.134499</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.210605</td>\n",
-       "      <td>0.209524</td>\n",
+       "      <td>0.190932</td>\n",
+       "      <td>0.079689</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.210605</td>\n",
-       "      <td>0.209524</td>\n",
+       "      <td>0.190932</td>\n",
+       "      <td>0.079689</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.124660</td>\n",
-       "      <td>0.119844</td>\n",
+       "      <td>0.114792</td>\n",
+       "      <td>0.047310</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.060003</td>\n",
-       "      <td>0.128876</td>\n",
+       "      <td>0.035533</td>\n",
+       "      <td>0.101868</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.057328</td>\n",
-       "      <td>0.040328</td>\n",
+       "      <td>0.055388</td>\n",
+       "      <td>0.041906</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.284902</td>\n",
-       "      <td>2.099200</td>\n",
+       "      <td>0.220158</td>\n",
+       "      <td>1.118408</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -717,14 +717,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.103547  0.049493\n",
-       "Predictive Parity                                0.074152  0.184299\n",
-       "Equal Opportunity                                0.210605  0.209524\n",
-       "Average Group Difference in False Negative Rate  0.210605  0.209524\n",
-       "Equalized Odds                                   0.124660  0.119844\n",
-       "Conditional Use Accuracy                         0.060003  0.128876\n",
-       "Average Group Difference in Accuracy             0.057328  0.040328\n",
-       "Treatment Equality                               0.284902  2.099200"
+       "Statistical Parity                               0.125058  0.073183\n",
+       "Predictive Parity                                0.013161  0.134499\n",
+       "Equal Opportunity                                0.190932  0.079689\n",
+       "Average Group Difference in False Negative Rate  0.190932  0.079689\n",
+       "Equalized Odds                                   0.114792  0.047310\n",
+       "Conditional Use Accuracy                         0.035533  0.101868\n",
+       "Average Group Difference in Accuracy             0.055388  0.041906\n",
+       "Treatment Equality                               0.220158  1.118408"
       ]
      },
      "execution_count": 10,
@@ -777,43 +777,43 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Statistical Parity</th>\n",
-       "      <td>0.093956</td>\n",
-       "      <td>0.019959</td>\n",
+       "      <td>0.095165</td>\n",
+       "      <td>0.016903</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Predictive Parity</th>\n",
-       "      <td>0.191814</td>\n",
-       "      <td>0.133510</td>\n",
+       "      <td>0.106777</td>\n",
+       "      <td>0.142254</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equal Opportunity</th>\n",
-       "      <td>0.145525</td>\n",
-       "      <td>0.191602</td>\n",
+       "      <td>0.191597</td>\n",
+       "      <td>0.177972</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in False Negative Rate</th>\n",
-       "      <td>0.145525</td>\n",
-       "      <td>0.191602</td>\n",
+       "      <td>0.191597</td>\n",
+       "      <td>0.177972</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Equalized Odds</th>\n",
-       "      <td>0.093755</td>\n",
-       "      <td>0.107181</td>\n",
+       "      <td>0.118433</td>\n",
+       "      <td>0.099354</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Conditional Use Accuracy</th>\n",
-       "      <td>0.115694</td>\n",
-       "      <td>0.104046</td>\n",
+       "      <td>0.070180</td>\n",
+       "      <td>0.100921</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.054694</td>\n",
-       "      <td>0.047595</td>\n",
+       "      <td>0.053687</td>\n",
+       "      <td>0.048322</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.292782</td>\n",
-       "      <td>2.399059</td>\n",
+       "      <td>0.394770</td>\n",
+       "      <td>1.736889</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -821,14 +821,14 @@
       ],
       "text/plain": [
        "                                                 original   updated\n",
-       "Statistical Parity                               0.093956  0.019959\n",
-       "Predictive Parity                                0.191814  0.133510\n",
-       "Equal Opportunity                                0.145525  0.191602\n",
-       "Average Group Difference in False Negative Rate  0.145525  0.191602\n",
-       "Equalized Odds                                   0.093755  0.107181\n",
-       "Conditional Use Accuracy                         0.115694  0.104046\n",
-       "Average Group Difference in Accuracy             0.054694  0.047595\n",
-       "Treatment Equality                               0.292782  2.399059"
+       "Statistical Parity                               0.095165  0.016903\n",
+       "Predictive Parity                                0.106777  0.142254\n",
+       "Equal Opportunity                                0.191597  0.177972\n",
+       "Average Group Difference in False Negative Rate  0.191597  0.177972\n",
+       "Equalized Odds                                   0.118433  0.099354\n",
+       "Conditional Use Accuracy                         0.070180  0.100921\n",
+       "Average Group Difference in Accuracy             0.053687  0.048322\n",
+       "Treatment Equality                               0.394770  1.736889"
       ]
      },
      "execution_count": 11,
@@ -854,7 +854,7 @@
    "outputs": [
     {
      "data": {
-      "image/png": "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",
+      "image/png": "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",
       "text/plain": [
        "<Figure size 640x480 with 1 Axes>"
       ]
@@ -882,7 +882,7 @@
    "outputs": [
     {
      "data": {
-      "image/png": "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",
+      "image/png": "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",
       "text/plain": [
        "<Figure size 640x480 with 1 Axes>"
       ]
@@ -962,200 +962,200 @@
        "    <tr>\n",
        "      <th rowspan=\"7\" valign=\"top\">original</th>\n",
        "      <th>Overall</th>\n",
-       "      <td>0.867988</td>\n",
-       "      <td>0.795825</td>\n",
-       "      <td>0.704437</td>\n",
-       "      <td>0.622647</td>\n",
-       "      <td>0.758689</td>\n",
-       "      <td>0.657426</td>\n",
-       "      <td>0.923695</td>\n",
+       "      <td>0.874212</td>\n",
+       "      <td>0.800385</td>\n",
+       "      <td>0.714816</td>\n",
+       "      <td>0.638670</td>\n",
+       "      <td>0.781250</td>\n",
+       "      <td>0.658795</td>\n",
+       "      <td>0.926794</td>\n",
        "      <td>2922.0</td>\n",
        "      <td>9289.0</td>\n",
        "      <td>0.239292</td>\n",
-       "      <td>0.207354</td>\n",
+       "      <td>0.201785</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
-       "      <td>0.892308</td>\n",
-       "      <td>0.589364</td>\n",
-       "      <td>0.300000</td>\n",
-       "      <td>0.340009</td>\n",
-       "      <td>0.750000</td>\n",
-       "      <td>0.187500</td>\n",
-       "      <td>0.877741</td>\n",
+       "      <td>0.863158</td>\n",
+       "      <td>0.618671</td>\n",
+       "      <td>0.380952</td>\n",
+       "      <td>0.397779</td>\n",
+       "      <td>0.800000</td>\n",
+       "      <td>0.250000</td>\n",
+       "      <td>0.879747</td>\n",
        "      <td>16.0</td>\n",
-       "      <td>114.0</td>\n",
-       "      <td>0.123077</td>\n",
-       "      <td>0.030769</td>\n",
+       "      <td>79.0</td>\n",
+       "      <td>0.168421</td>\n",
+       "      <td>0.052632</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
-       "      <td>0.831978</td>\n",
-       "      <td>0.754766</td>\n",
-       "      <td>0.655556</td>\n",
-       "      <td>0.553843</td>\n",
-       "      <td>0.746835</td>\n",
-       "      <td>0.584158</td>\n",
-       "      <td>0.908711</td>\n",
-       "      <td>101.0</td>\n",
-       "      <td>268.0</td>\n",
-       "      <td>0.273713</td>\n",
-       "      <td>0.214092</td>\n",
+       "      <td>0.837838</td>\n",
+       "      <td>0.805269</td>\n",
+       "      <td>0.750000</td>\n",
+       "      <td>0.634899</td>\n",
+       "      <td>0.810811</td>\n",
+       "      <td>0.697674</td>\n",
+       "      <td>0.919554</td>\n",
+       "      <td>129.0</td>\n",
+       "      <td>241.0</td>\n",
+       "      <td>0.348649</td>\n",
+       "      <td>0.300000</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
-       "      <td>0.929614</td>\n",
-       "      <td>0.767418</td>\n",
-       "      <td>0.658333</td>\n",
-       "      <td>0.635246</td>\n",
-       "      <td>0.814433</td>\n",
-       "      <td>0.552448</td>\n",
-       "      <td>0.947019</td>\n",
-       "      <td>143.0</td>\n",
-       "      <td>1022.0</td>\n",
-       "      <td>0.122747</td>\n",
-       "      <td>0.083262</td>\n",
+       "      <td>0.939138</td>\n",
+       "      <td>0.792021</td>\n",
+       "      <td>0.686957</td>\n",
+       "      <td>0.661844</td>\n",
+       "      <td>0.797980</td>\n",
+       "      <td>0.603053</td>\n",
+       "      <td>0.960479</td>\n",
+       "      <td>131.0</td>\n",
+       "      <td>1052.0</td>\n",
+       "      <td>0.110735</td>\n",
+       "      <td>0.083686</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
-       "      <td>0.941176</td>\n",
-       "      <td>0.815747</td>\n",
-       "      <td>0.750000</td>\n",
-       "      <td>0.730769</td>\n",
-       "      <td>0.900000</td>\n",
-       "      <td>0.642857</td>\n",
-       "      <td>0.956169</td>\n",
-       "      <td>14.0</td>\n",
-       "      <td>88.0</td>\n",
-       "      <td>0.137255</td>\n",
-       "      <td>0.098039</td>\n",
+       "      <td>0.937500</td>\n",
+       "      <td>0.797591</td>\n",
+       "      <td>0.695652</td>\n",
+       "      <td>0.668558</td>\n",
+       "      <td>0.800000</td>\n",
+       "      <td>0.615385</td>\n",
+       "      <td>0.973582</td>\n",
+       "      <td>13.0</td>\n",
+       "      <td>99.0</td>\n",
+       "      <td>0.116071</td>\n",
+       "      <td>0.089286</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
-       "      <td>0.861369</td>\n",
-       "      <td>0.797787</td>\n",
-       "      <td>0.709820</td>\n",
-       "      <td>0.621215</td>\n",
-       "      <td>0.756191</td>\n",
-       "      <td>0.668807</td>\n",
-       "      <td>0.921111</td>\n",
-       "      <td>2648.0</td>\n",
-       "      <td>7797.0</td>\n",
-       "      <td>0.253518</td>\n",
-       "      <td>0.224222</td>\n",
+       "      <td>0.867572</td>\n",
+       "      <td>0.799523</td>\n",
+       "      <td>0.715928</td>\n",
+       "      <td>0.633844</td>\n",
+       "      <td>0.778919</td>\n",
+       "      <td>0.662362</td>\n",
+       "      <td>0.922501</td>\n",
+       "      <td>2633.0</td>\n",
+       "      <td>7818.0</td>\n",
+       "      <td>0.251938</td>\n",
+       "      <td>0.214238</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Maximum difference</th>\n",
-       "      <td>0.109198</td>\n",
-       "      <td>0.226383</td>\n",
-       "      <td>0.450000</td>\n",
-       "      <td>0.390761</td>\n",
-       "      <td>0.153165</td>\n",
-       "      <td>0.481307</td>\n",
-       "      <td>0.078428</td>\n",
-       "      <td>2634.0</td>\n",
-       "      <td>7709.0</td>\n",
-       "      <td>0.150966</td>\n",
-       "      <td>0.193453</td>\n",
+       "      <td>0.101300</td>\n",
+       "      <td>0.186598</td>\n",
+       "      <td>0.369048</td>\n",
+       "      <td>0.270780</td>\n",
+       "      <td>0.031892</td>\n",
+       "      <td>0.447674</td>\n",
+       "      <td>0.093835</td>\n",
+       "      <td>2620.0</td>\n",
+       "      <td>7739.0</td>\n",
+       "      <td>0.237913</td>\n",
+       "      <td>0.247368</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th rowspan=\"7\" valign=\"top\">updated</th>\n",
        "      <th>Overall</th>\n",
-       "      <td>0.864057</td>\n",
-       "      <td>0.769549</td>\n",
-       "      <td>0.674382</td>\n",
-       "      <td>0.601041</td>\n",
-       "      <td>0.789982</td>\n",
-       "      <td>0.588296</td>\n",
-       "      <td>0.900496</td>\n",
+       "      <td>0.872001</td>\n",
+       "      <td>0.792246</td>\n",
+       "      <td>0.705039</td>\n",
+       "      <td>0.629830</td>\n",
+       "      <td>0.785865</td>\n",
+       "      <td>0.639288</td>\n",
+       "      <td>0.907573</td>\n",
        "      <td>2922.0</td>\n",
        "      <td>9289.0</td>\n",
        "      <td>0.239292</td>\n",
-       "      <td>0.178200</td>\n",
+       "      <td>0.194661</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
-       "      <td>0.869231</td>\n",
-       "      <td>0.683662</td>\n",
-       "      <td>0.451613</td>\n",
-       "      <td>0.377718</td>\n",
-       "      <td>0.466667</td>\n",
-       "      <td>0.437500</td>\n",
-       "      <td>0.877741</td>\n",
+       "      <td>0.884211</td>\n",
+       "      <td>0.780854</td>\n",
+       "      <td>0.645161</td>\n",
+       "      <td>0.576493</td>\n",
+       "      <td>0.666667</td>\n",
+       "      <td>0.625000</td>\n",
+       "      <td>0.879747</td>\n",
        "      <td>16.0</td>\n",
-       "      <td>114.0</td>\n",
-       "      <td>0.123077</td>\n",
-       "      <td>0.115385</td>\n",
+       "      <td>79.0</td>\n",
+       "      <td>0.168421</td>\n",
+       "      <td>0.157895</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
-       "      <td>0.826558</td>\n",
-       "      <td>0.726356</td>\n",
-       "      <td>0.614458</td>\n",
-       "      <td>0.529854</td>\n",
-       "      <td>0.784615</td>\n",
-       "      <td>0.504950</td>\n",
-       "      <td>0.908711</td>\n",
-       "      <td>101.0</td>\n",
-       "      <td>268.0</td>\n",
-       "      <td>0.273713</td>\n",
-       "      <td>0.176152</td>\n",
+       "      <td>0.837838</td>\n",
+       "      <td>0.794461</td>\n",
+       "      <td>0.736842</td>\n",
+       "      <td>0.633956</td>\n",
+       "      <td>0.848485</td>\n",
+       "      <td>0.651163</td>\n",
+       "      <td>0.919554</td>\n",
+       "      <td>129.0</td>\n",
+       "      <td>241.0</td>\n",
+       "      <td>0.348649</td>\n",
+       "      <td>0.267568</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
-       "      <td>0.915880</td>\n",
-       "      <td>0.828757</td>\n",
-       "      <td>0.675497</td>\n",
-       "      <td>0.628493</td>\n",
-       "      <td>0.641509</td>\n",
-       "      <td>0.713287</td>\n",
-       "      <td>0.947019</td>\n",
-       "      <td>143.0</td>\n",
-       "      <td>1022.0</td>\n",
-       "      <td>0.122747</td>\n",
-       "      <td>0.136481</td>\n",
+       "      <td>0.913779</td>\n",
+       "      <td>0.864642</td>\n",
+       "      <td>0.673077</td>\n",
+       "      <td>0.635721</td>\n",
+       "      <td>0.580110</td>\n",
+       "      <td>0.801527</td>\n",
+       "      <td>0.960479</td>\n",
+       "      <td>131.0</td>\n",
+       "      <td>1052.0</td>\n",
+       "      <td>0.110735</td>\n",
+       "      <td>0.153001</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
-       "      <td>0.882353</td>\n",
-       "      <td>0.871753</td>\n",
+       "      <td>0.919643</td>\n",
+       "      <td>0.787490</td>\n",
+       "      <td>0.640000</td>\n",
+       "      <td>0.595459</td>\n",
        "      <td>0.666667</td>\n",
-       "      <td>0.622062</td>\n",
-       "      <td>0.545455</td>\n",
-       "      <td>0.857143</td>\n",
-       "      <td>0.956169</td>\n",
-       "      <td>14.0</td>\n",
-       "      <td>88.0</td>\n",
-       "      <td>0.137255</td>\n",
-       "      <td>0.215686</td>\n",
+       "      <td>0.615385</td>\n",
+       "      <td>0.973582</td>\n",
+       "      <td>13.0</td>\n",
+       "      <td>99.0</td>\n",
+       "      <td>0.116071</td>\n",
+       "      <td>0.107143</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
-       "      <td>0.859359</td>\n",
-       "      <td>0.768508</td>\n",
-       "      <td>0.678063</td>\n",
-       "      <td>0.603743</td>\n",
-       "      <td>0.807833</td>\n",
-       "      <td>0.584215</td>\n",
-       "      <td>0.921111</td>\n",
-       "      <td>2648.0</td>\n",
-       "      <td>7797.0</td>\n",
-       "      <td>0.253518</td>\n",
-       "      <td>0.183341</td>\n",
+       "      <td>0.867860</td>\n",
+       "      <td>0.789262</td>\n",
+       "      <td>0.706358</td>\n",
+       "      <td>0.630176</td>\n",
+       "      <td>0.802415</td>\n",
+       "      <td>0.630839</td>\n",
+       "      <td>0.922501</td>\n",
+       "      <td>2633.0</td>\n",
+       "      <td>7818.0</td>\n",
+       "      <td>0.251938</td>\n",
+       "      <td>0.198067</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Maximum difference</th>\n",
-       "      <td>0.089322</td>\n",
-       "      <td>0.188091</td>\n",
-       "      <td>0.226450</td>\n",
-       "      <td>0.250775</td>\n",
-       "      <td>0.341166</td>\n",
-       "      <td>0.419643</td>\n",
-       "      <td>0.078428</td>\n",
-       "      <td>2634.0</td>\n",
-       "      <td>7709.0</td>\n",
-       "      <td>0.150966</td>\n",
-       "      <td>0.100302</td>\n",
+       "      <td>0.081805</td>\n",
+       "      <td>0.083787</td>\n",
+       "      <td>0.096842</td>\n",
+       "      <td>0.059228</td>\n",
+       "      <td>0.268374</td>\n",
+       "      <td>0.186142</td>\n",
+       "      <td>0.093835</td>\n",
+       "      <td>2620.0</td>\n",
+       "      <td>7739.0</td>\n",
+       "      <td>0.237913</td>\n",
+       "      <td>0.160425</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1164,71 +1164,71 @@
       "text/plain": [
        "                             Accuracy  Balanced Accuracy  F1 score       MCC  \\\n",
        "         Groups                                                                \n",
-       "original Overall             0.867988           0.795825  0.704437  0.622647   \n",
-       "         0                   0.892308           0.589364  0.300000  0.340009   \n",
-       "         1                   0.831978           0.754766  0.655556  0.553843   \n",
-       "         2                   0.929614           0.767418  0.658333  0.635246   \n",
-       "         3                   0.941176           0.815747  0.750000  0.730769   \n",
-       "         4                   0.861369           0.797787  0.709820  0.621215   \n",
-       "         Maximum difference  0.109198           0.226383  0.450000  0.390761   \n",
-       "updated  Overall             0.864057           0.769549  0.674382  0.601041   \n",
-       "         0                   0.869231           0.683662  0.451613  0.377718   \n",
-       "         1                   0.826558           0.726356  0.614458  0.529854   \n",
-       "         2                   0.915880           0.828757  0.675497  0.628493   \n",
-       "         3                   0.882353           0.871753  0.666667  0.622062   \n",
-       "         4                   0.859359           0.768508  0.678063  0.603743   \n",
-       "         Maximum difference  0.089322           0.188091  0.226450  0.250775   \n",
+       "original Overall             0.874212           0.800385  0.714816  0.638670   \n",
+       "         0                   0.863158           0.618671  0.380952  0.397779   \n",
+       "         1                   0.837838           0.805269  0.750000  0.634899   \n",
+       "         2                   0.939138           0.792021  0.686957  0.661844   \n",
+       "         3                   0.937500           0.797591  0.695652  0.668558   \n",
+       "         4                   0.867572           0.799523  0.715928  0.633844   \n",
+       "         Maximum difference  0.101300           0.186598  0.369048  0.270780   \n",
+       "updated  Overall             0.872001           0.792246  0.705039  0.629830   \n",
+       "         0                   0.884211           0.780854  0.645161  0.576493   \n",
+       "         1                   0.837838           0.794461  0.736842  0.633956   \n",
+       "         2                   0.913779           0.864642  0.673077  0.635721   \n",
+       "         3                   0.919643           0.787490  0.640000  0.595459   \n",
+       "         4                   0.867860           0.789262  0.706358  0.630176   \n",
+       "         Maximum difference  0.081805           0.083787  0.096842  0.059228   \n",
        "\n",
        "                             Precision    Recall   ROC AUC  Positive Count  \\\n",
        "         Groups                                                              \n",
-       "original Overall              0.758689  0.657426  0.923695          2922.0   \n",
-       "         0                    0.750000  0.187500  0.877741            16.0   \n",
-       "         1                    0.746835  0.584158  0.908711           101.0   \n",
-       "         2                    0.814433  0.552448  0.947019           143.0   \n",
-       "         3                    0.900000  0.642857  0.956169            14.0   \n",
-       "         4                    0.756191  0.668807  0.921111          2648.0   \n",
-       "         Maximum difference   0.153165  0.481307  0.078428          2634.0   \n",
-       "updated  Overall              0.789982  0.588296  0.900496          2922.0   \n",
-       "         0                    0.466667  0.437500  0.877741            16.0   \n",
-       "         1                    0.784615  0.504950  0.908711           101.0   \n",
-       "         2                    0.641509  0.713287  0.947019           143.0   \n",
-       "         3                    0.545455  0.857143  0.956169            14.0   \n",
-       "         4                    0.807833  0.584215  0.921111          2648.0   \n",
-       "         Maximum difference   0.341166  0.419643  0.078428          2634.0   \n",
+       "original Overall              0.781250  0.658795  0.926794          2922.0   \n",
+       "         0                    0.800000  0.250000  0.879747            16.0   \n",
+       "         1                    0.810811  0.697674  0.919554           129.0   \n",
+       "         2                    0.797980  0.603053  0.960479           131.0   \n",
+       "         3                    0.800000  0.615385  0.973582            13.0   \n",
+       "         4                    0.778919  0.662362  0.922501          2633.0   \n",
+       "         Maximum difference   0.031892  0.447674  0.093835          2620.0   \n",
+       "updated  Overall              0.785865  0.639288  0.907573          2922.0   \n",
+       "         0                    0.666667  0.625000  0.879747            16.0   \n",
+       "         1                    0.848485  0.651163  0.919554           129.0   \n",
+       "         2                    0.580110  0.801527  0.960479           131.0   \n",
+       "         3                    0.666667  0.615385  0.973582            13.0   \n",
+       "         4                    0.802415  0.630839  0.922501          2633.0   \n",
+       "         Maximum difference   0.268374  0.186142  0.093835          2620.0   \n",
        "\n",
        "                             Negative Count  Positive Label Rate  \\\n",
        "         Groups                                                    \n",
        "original Overall                     9289.0             0.239292   \n",
-       "         0                            114.0             0.123077   \n",
-       "         1                            268.0             0.273713   \n",
-       "         2                           1022.0             0.122747   \n",
-       "         3                             88.0             0.137255   \n",
-       "         4                           7797.0             0.253518   \n",
-       "         Maximum difference          7709.0             0.150966   \n",
+       "         0                             79.0             0.168421   \n",
+       "         1                            241.0             0.348649   \n",
+       "         2                           1052.0             0.110735   \n",
+       "         3                             99.0             0.116071   \n",
+       "         4                           7818.0             0.251938   \n",
+       "         Maximum difference          7739.0             0.237913   \n",
        "updated  Overall                     9289.0             0.239292   \n",
-       "         0                            114.0             0.123077   \n",
-       "         1                            268.0             0.273713   \n",
-       "         2                           1022.0             0.122747   \n",
-       "         3                             88.0             0.137255   \n",
-       "         4                           7797.0             0.253518   \n",
-       "         Maximum difference          7709.0             0.150966   \n",
+       "         0                             79.0             0.168421   \n",
+       "         1                            241.0             0.348649   \n",
+       "         2                           1052.0             0.110735   \n",
+       "         3                             99.0             0.116071   \n",
+       "         4                           7818.0             0.251938   \n",
+       "         Maximum difference          7739.0             0.237913   \n",
        "\n",
        "                             Positive Prediction Rate  \n",
        "         Groups                                        \n",
-       "original Overall                             0.207354  \n",
-       "         0                                   0.030769  \n",
-       "         1                                   0.214092  \n",
-       "         2                                   0.083262  \n",
-       "         3                                   0.098039  \n",
-       "         4                                   0.224222  \n",
-       "         Maximum difference                  0.193453  \n",
-       "updated  Overall                             0.178200  \n",
-       "         0                                   0.115385  \n",
-       "         1                                   0.176152  \n",
-       "         2                                   0.136481  \n",
-       "         3                                   0.215686  \n",
-       "         4                                   0.183341  \n",
-       "         Maximum difference                  0.100302  "
+       "original Overall                             0.201785  \n",
+       "         0                                   0.052632  \n",
+       "         1                                   0.300000  \n",
+       "         2                                   0.083686  \n",
+       "         3                                   0.089286  \n",
+       "         4                                   0.214238  \n",
+       "         Maximum difference                  0.247368  \n",
+       "updated  Overall                             0.194661  \n",
+       "         0                                   0.157895  \n",
+       "         1                                   0.267568  \n",
+       "         2                                   0.153001  \n",
+       "         3                                   0.107143  \n",
+       "         4                                   0.198067  \n",
+       "         Maximum difference                  0.160425  "
       ]
      },
      "execution_count": 14,
@@ -1308,200 +1308,200 @@
        "    <tr>\n",
        "      <th rowspan=\"7\" valign=\"top\">original</th>\n",
        "      <th>Overall</th>\n",
-       "      <td>0.866749</td>\n",
-       "      <td>0.790909</td>\n",
-       "      <td>0.698648</td>\n",
-       "      <td>0.617308</td>\n",
-       "      <td>0.761405</td>\n",
-       "      <td>0.645448</td>\n",
-       "      <td>0.920445</td>\n",
+       "      <td>0.873464</td>\n",
+       "      <td>0.800249</td>\n",
+       "      <td>0.713942</td>\n",
+       "      <td>0.636936</td>\n",
+       "      <td>0.777733</td>\n",
+       "      <td>0.659822</td>\n",
+       "      <td>0.928441</td>\n",
        "      <td>2922.0</td>\n",
        "      <td>9288.0</td>\n",
        "      <td>0.239312</td>\n",
-       "      <td>0.202867</td>\n",
+       "      <td>0.203030</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
-       "      <td>0.883929</td>\n",
-       "      <td>0.658163</td>\n",
-       "      <td>0.434783</td>\n",
-       "      <td>0.384833</td>\n",
-       "      <td>0.555556</td>\n",
-       "      <td>0.357143</td>\n",
-       "      <td>0.948251</td>\n",
-       "      <td>14.0</td>\n",
-       "      <td>98.0</td>\n",
-       "      <td>0.125000</td>\n",
-       "      <td>0.080357</td>\n",
+       "      <td>0.888889</td>\n",
+       "      <td>0.615909</td>\n",
+       "      <td>0.363636</td>\n",
+       "      <td>0.362441</td>\n",
+       "      <td>0.666667</td>\n",
+       "      <td>0.250000</td>\n",
+       "      <td>0.900000</td>\n",
+       "      <td>16.0</td>\n",
+       "      <td>110.0</td>\n",
+       "      <td>0.126984</td>\n",
+       "      <td>0.047619</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
-       "      <td>0.836412</td>\n",
-       "      <td>0.756772</td>\n",
-       "      <td>0.651685</td>\n",
-       "      <td>0.551016</td>\n",
-       "      <td>0.725000</td>\n",
-       "      <td>0.591837</td>\n",
-       "      <td>0.893093</td>\n",
-       "      <td>98.0</td>\n",
-       "      <td>281.0</td>\n",
-       "      <td>0.258575</td>\n",
-       "      <td>0.211082</td>\n",
+       "      <td>0.831978</td>\n",
+       "      <td>0.759259</td>\n",
+       "      <td>0.621951</td>\n",
+       "      <td>0.514021</td>\n",
+       "      <td>0.614458</td>\n",
+       "      <td>0.629630</td>\n",
+       "      <td>0.879630</td>\n",
+       "      <td>81.0</td>\n",
+       "      <td>288.0</td>\n",
+       "      <td>0.219512</td>\n",
+       "      <td>0.224932</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
-       "      <td>0.931217</td>\n",
-       "      <td>0.751402</td>\n",
-       "      <td>0.628571</td>\n",
-       "      <td>0.608799</td>\n",
-       "      <td>0.795181</td>\n",
-       "      <td>0.519685</td>\n",
-       "      <td>0.953139</td>\n",
-       "      <td>127.0</td>\n",
-       "      <td>1007.0</td>\n",
-       "      <td>0.111993</td>\n",
-       "      <td>0.073192</td>\n",
+       "      <td>0.927920</td>\n",
+       "      <td>0.785367</td>\n",
+       "      <td>0.678967</td>\n",
+       "      <td>0.647838</td>\n",
+       "      <td>0.793103</td>\n",
+       "      <td>0.593548</td>\n",
+       "      <td>0.947179</td>\n",
+       "      <td>155.0</td>\n",
+       "      <td>1052.0</td>\n",
+       "      <td>0.128418</td>\n",
+       "      <td>0.096106</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
-       "      <td>0.937500</td>\n",
-       "      <td>0.730769</td>\n",
-       "      <td>0.631579</td>\n",
-       "      <td>0.656551</td>\n",
-       "      <td>1.000000</td>\n",
-       "      <td>0.461538</td>\n",
-       "      <td>0.963481</td>\n",
+       "      <td>0.936170</td>\n",
+       "      <td>0.833808</td>\n",
+       "      <td>0.750000</td>\n",
+       "      <td>0.716978</td>\n",
+       "      <td>0.818182</td>\n",
+       "      <td>0.692308</td>\n",
+       "      <td>0.951567</td>\n",
        "      <td>13.0</td>\n",
-       "      <td>99.0</td>\n",
-       "      <td>0.116071</td>\n",
-       "      <td>0.053571</td>\n",
+       "      <td>81.0</td>\n",
+       "      <td>0.138298</td>\n",
+       "      <td>0.117021</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
-       "      <td>0.859926</td>\n",
-       "      <td>0.792788</td>\n",
-       "      <td>0.704770</td>\n",
-       "      <td>0.616571</td>\n",
-       "      <td>0.761635</td>\n",
-       "      <td>0.655805</td>\n",
-       "      <td>0.916189</td>\n",
-       "      <td>2670.0</td>\n",
-       "      <td>7803.0</td>\n",
-       "      <td>0.254941</td>\n",
-       "      <td>0.219517</td>\n",
+       "      <td>0.867870</td>\n",
+       "      <td>0.801810</td>\n",
+       "      <td>0.720325</td>\n",
+       "      <td>0.638056</td>\n",
+       "      <td>0.783031</td>\n",
+       "      <td>0.666918</td>\n",
+       "      <td>0.926850</td>\n",
+       "      <td>2657.0</td>\n",
+       "      <td>7757.0</td>\n",
+       "      <td>0.255137</td>\n",
+       "      <td>0.217304</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Maximum difference</th>\n",
-       "      <td>0.101088</td>\n",
-       "      <td>0.134625</td>\n",
-       "      <td>0.269987</td>\n",
-       "      <td>0.271718</td>\n",
-       "      <td>0.444444</td>\n",
-       "      <td>0.298662</td>\n",
-       "      <td>0.070388</td>\n",
-       "      <td>2657.0</td>\n",
-       "      <td>7705.0</td>\n",
-       "      <td>0.146582</td>\n",
-       "      <td>0.165945</td>\n",
+       "      <td>0.104192</td>\n",
+       "      <td>0.217899</td>\n",
+       "      <td>0.386364</td>\n",
+       "      <td>0.354537</td>\n",
+       "      <td>0.203724</td>\n",
+       "      <td>0.442308</td>\n",
+       "      <td>0.071937</td>\n",
+       "      <td>2644.0</td>\n",
+       "      <td>7676.0</td>\n",
+       "      <td>0.128153</td>\n",
+       "      <td>0.177313</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th rowspan=\"7\" valign=\"top\">updated</th>\n",
        "      <th>Overall</th>\n",
-       "      <td>0.864537</td>\n",
-       "      <td>0.770221</td>\n",
-       "      <td>0.675559</td>\n",
-       "      <td>0.602537</td>\n",
-       "      <td>0.791360</td>\n",
-       "      <td>0.589322</td>\n",
-       "      <td>0.894418</td>\n",
+       "      <td>0.872072</td>\n",
+       "      <td>0.791593</td>\n",
+       "      <td>0.704502</td>\n",
+       "      <td>0.629731</td>\n",
+       "      <td>0.787648</td>\n",
+       "      <td>0.637235</td>\n",
+       "      <td>0.908668</td>\n",
        "      <td>2922.0</td>\n",
        "      <td>9288.0</td>\n",
        "      <td>0.239312</td>\n",
-       "      <td>0.178215</td>\n",
+       "      <td>0.193612</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
-       "      <td>0.910714</td>\n",
-       "      <td>0.887755</td>\n",
-       "      <td>0.705882</td>\n",
-       "      <td>0.669662</td>\n",
-       "      <td>0.600000</td>\n",
-       "      <td>0.857143</td>\n",
-       "      <td>0.948251</td>\n",
-       "      <td>14.0</td>\n",
-       "      <td>98.0</td>\n",
-       "      <td>0.125000</td>\n",
-       "      <td>0.178571</td>\n",
+       "      <td>0.880952</td>\n",
+       "      <td>0.798295</td>\n",
+       "      <td>0.594595</td>\n",
+       "      <td>0.533002</td>\n",
+       "      <td>0.523810</td>\n",
+       "      <td>0.687500</td>\n",
+       "      <td>0.900000</td>\n",
+       "      <td>16.0</td>\n",
+       "      <td>110.0</td>\n",
+       "      <td>0.126984</td>\n",
+       "      <td>0.166667</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
-       "      <td>0.828496</td>\n",
-       "      <td>0.724853</td>\n",
-       "      <td>0.606061</td>\n",
-       "      <td>0.516155</td>\n",
-       "      <td>0.746269</td>\n",
-       "      <td>0.510204</td>\n",
-       "      <td>0.893093</td>\n",
-       "      <td>98.0</td>\n",
-       "      <td>281.0</td>\n",
-       "      <td>0.258575</td>\n",
-       "      <td>0.176781</td>\n",
+       "      <td>0.845528</td>\n",
+       "      <td>0.736883</td>\n",
+       "      <td>0.606897</td>\n",
+       "      <td>0.517919</td>\n",
+       "      <td>0.687500</td>\n",
+       "      <td>0.543210</td>\n",
+       "      <td>0.879630</td>\n",
+       "      <td>81.0</td>\n",
+       "      <td>288.0</td>\n",
+       "      <td>0.219512</td>\n",
+       "      <td>0.173442</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
-       "      <td>0.921517</td>\n",
-       "      <td>0.845714</td>\n",
-       "      <td>0.681004</td>\n",
-       "      <td>0.640009</td>\n",
-       "      <td>0.625000</td>\n",
-       "      <td>0.748031</td>\n",
-       "      <td>0.953139</td>\n",
-       "      <td>127.0</td>\n",
-       "      <td>1007.0</td>\n",
-       "      <td>0.111993</td>\n",
-       "      <td>0.134039</td>\n",
+       "      <td>0.907208</td>\n",
+       "      <td>0.844999</td>\n",
+       "      <td>0.678161</td>\n",
+       "      <td>0.629830</td>\n",
+       "      <td>0.611399</td>\n",
+       "      <td>0.761290</td>\n",
+       "      <td>0.947179</td>\n",
+       "      <td>155.0</td>\n",
+       "      <td>1052.0</td>\n",
+       "      <td>0.128418</td>\n",
+       "      <td>0.159901</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
-       "      <td>0.901786</td>\n",
-       "      <td>0.877622</td>\n",
-       "      <td>0.666667</td>\n",
-       "      <td>0.631638</td>\n",
-       "      <td>0.550000</td>\n",
-       "      <td>0.846154</td>\n",
-       "      <td>0.963481</td>\n",
+       "      <td>0.946809</td>\n",
+       "      <td>0.936847</td>\n",
+       "      <td>0.827586</td>\n",
+       "      <td>0.802539</td>\n",
+       "      <td>0.750000</td>\n",
+       "      <td>0.923077</td>\n",
+       "      <td>0.951567</td>\n",
        "      <td>13.0</td>\n",
-       "      <td>99.0</td>\n",
-       "      <td>0.116071</td>\n",
-       "      <td>0.178571</td>\n",
+       "      <td>81.0</td>\n",
+       "      <td>0.138298</td>\n",
+       "      <td>0.170213</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
-       "      <td>0.858780</td>\n",
-       "      <td>0.767751</td>\n",
-       "      <td>0.677567</td>\n",
-       "      <td>0.603533</td>\n",
-       "      <td>0.810642</td>\n",
-       "      <td>0.582022</td>\n",
-       "      <td>0.916189</td>\n",
-       "      <td>2670.0</td>\n",
-       "      <td>7803.0</td>\n",
-       "      <td>0.254941</td>\n",
-       "      <td>0.183042</td>\n",
+       "      <td>0.868158</td>\n",
+       "      <td>0.790250</td>\n",
+       "      <td>0.709541</td>\n",
+       "      <td>0.634121</td>\n",
+       "      <td>0.810145</td>\n",
+       "      <td>0.631163</td>\n",
+       "      <td>0.926850</td>\n",
+       "      <td>2657.0</td>\n",
+       "      <td>7757.0</td>\n",
+       "      <td>0.255137</td>\n",
+       "      <td>0.198771</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Maximum difference</th>\n",
-       "      <td>0.093021</td>\n",
-       "      <td>0.162902</td>\n",
-       "      <td>0.099822</td>\n",
-       "      <td>0.153507</td>\n",
-       "      <td>0.260642</td>\n",
-       "      <td>0.346939</td>\n",
-       "      <td>0.070388</td>\n",
-       "      <td>2657.0</td>\n",
-       "      <td>7705.0</td>\n",
-       "      <td>0.146582</td>\n",
-       "      <td>0.049003</td>\n",
+       "      <td>0.101280</td>\n",
+       "      <td>0.199964</td>\n",
+       "      <td>0.232992</td>\n",
+       "      <td>0.284620</td>\n",
+       "      <td>0.286335</td>\n",
+       "      <td>0.379867</td>\n",
+       "      <td>0.071937</td>\n",
+       "      <td>2644.0</td>\n",
+       "      <td>7676.0</td>\n",
+       "      <td>0.128153</td>\n",
+       "      <td>0.038870</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1510,71 +1510,71 @@
       "text/plain": [
        "                             Accuracy  Balanced Accuracy  F1 score       MCC  \\\n",
        "         Groups                                                                \n",
-       "original Overall             0.866749           0.790909  0.698648  0.617308   \n",
-       "         0                   0.883929           0.658163  0.434783  0.384833   \n",
-       "         1                   0.836412           0.756772  0.651685  0.551016   \n",
-       "         2                   0.931217           0.751402  0.628571  0.608799   \n",
-       "         3                   0.937500           0.730769  0.631579  0.656551   \n",
-       "         4                   0.859926           0.792788  0.704770  0.616571   \n",
-       "         Maximum difference  0.101088           0.134625  0.269987  0.271718   \n",
-       "updated  Overall             0.864537           0.770221  0.675559  0.602537   \n",
-       "         0                   0.910714           0.887755  0.705882  0.669662   \n",
-       "         1                   0.828496           0.724853  0.606061  0.516155   \n",
-       "         2                   0.921517           0.845714  0.681004  0.640009   \n",
-       "         3                   0.901786           0.877622  0.666667  0.631638   \n",
-       "         4                   0.858780           0.767751  0.677567  0.603533   \n",
-       "         Maximum difference  0.093021           0.162902  0.099822  0.153507   \n",
+       "original Overall             0.873464           0.800249  0.713942  0.636936   \n",
+       "         0                   0.888889           0.615909  0.363636  0.362441   \n",
+       "         1                   0.831978           0.759259  0.621951  0.514021   \n",
+       "         2                   0.927920           0.785367  0.678967  0.647838   \n",
+       "         3                   0.936170           0.833808  0.750000  0.716978   \n",
+       "         4                   0.867870           0.801810  0.720325  0.638056   \n",
+       "         Maximum difference  0.104192           0.217899  0.386364  0.354537   \n",
+       "updated  Overall             0.872072           0.791593  0.704502  0.629731   \n",
+       "         0                   0.880952           0.798295  0.594595  0.533002   \n",
+       "         1                   0.845528           0.736883  0.606897  0.517919   \n",
+       "         2                   0.907208           0.844999  0.678161  0.629830   \n",
+       "         3                   0.946809           0.936847  0.827586  0.802539   \n",
+       "         4                   0.868158           0.790250  0.709541  0.634121   \n",
+       "         Maximum difference  0.101280           0.199964  0.232992  0.284620   \n",
        "\n",
        "                             Precision    Recall   ROC AUC  Positive Count  \\\n",
        "         Groups                                                              \n",
-       "original Overall              0.761405  0.645448  0.920445          2922.0   \n",
-       "         0                    0.555556  0.357143  0.948251            14.0   \n",
-       "         1                    0.725000  0.591837  0.893093            98.0   \n",
-       "         2                    0.795181  0.519685  0.953139           127.0   \n",
-       "         3                    1.000000  0.461538  0.963481            13.0   \n",
-       "         4                    0.761635  0.655805  0.916189          2670.0   \n",
-       "         Maximum difference   0.444444  0.298662  0.070388          2657.0   \n",
-       "updated  Overall              0.791360  0.589322  0.894418          2922.0   \n",
-       "         0                    0.600000  0.857143  0.948251            14.0   \n",
-       "         1                    0.746269  0.510204  0.893093            98.0   \n",
-       "         2                    0.625000  0.748031  0.953139           127.0   \n",
-       "         3                    0.550000  0.846154  0.963481            13.0   \n",
-       "         4                    0.810642  0.582022  0.916189          2670.0   \n",
-       "         Maximum difference   0.260642  0.346939  0.070388          2657.0   \n",
+       "original Overall              0.777733  0.659822  0.928441          2922.0   \n",
+       "         0                    0.666667  0.250000  0.900000            16.0   \n",
+       "         1                    0.614458  0.629630  0.879630            81.0   \n",
+       "         2                    0.793103  0.593548  0.947179           155.0   \n",
+       "         3                    0.818182  0.692308  0.951567            13.0   \n",
+       "         4                    0.783031  0.666918  0.926850          2657.0   \n",
+       "         Maximum difference   0.203724  0.442308  0.071937          2644.0   \n",
+       "updated  Overall              0.787648  0.637235  0.908668          2922.0   \n",
+       "         0                    0.523810  0.687500  0.900000            16.0   \n",
+       "         1                    0.687500  0.543210  0.879630            81.0   \n",
+       "         2                    0.611399  0.761290  0.947179           155.0   \n",
+       "         3                    0.750000  0.923077  0.951567            13.0   \n",
+       "         4                    0.810145  0.631163  0.926850          2657.0   \n",
+       "         Maximum difference   0.286335  0.379867  0.071937          2644.0   \n",
        "\n",
        "                             Negative Count  Positive Label Rate  \\\n",
        "         Groups                                                    \n",
        "original Overall                     9288.0             0.239312   \n",
-       "         0                             98.0             0.125000   \n",
-       "         1                            281.0             0.258575   \n",
-       "         2                           1007.0             0.111993   \n",
-       "         3                             99.0             0.116071   \n",
-       "         4                           7803.0             0.254941   \n",
-       "         Maximum difference          7705.0             0.146582   \n",
+       "         0                            110.0             0.126984   \n",
+       "         1                            288.0             0.219512   \n",
+       "         2                           1052.0             0.128418   \n",
+       "         3                             81.0             0.138298   \n",
+       "         4                           7757.0             0.255137   \n",
+       "         Maximum difference          7676.0             0.128153   \n",
        "updated  Overall                     9288.0             0.239312   \n",
-       "         0                             98.0             0.125000   \n",
-       "         1                            281.0             0.258575   \n",
-       "         2                           1007.0             0.111993   \n",
-       "         3                             99.0             0.116071   \n",
-       "         4                           7803.0             0.254941   \n",
-       "         Maximum difference          7705.0             0.146582   \n",
+       "         0                            110.0             0.126984   \n",
+       "         1                            288.0             0.219512   \n",
+       "         2                           1052.0             0.128418   \n",
+       "         3                             81.0             0.138298   \n",
+       "         4                           7757.0             0.255137   \n",
+       "         Maximum difference          7676.0             0.128153   \n",
        "\n",
        "                             Positive Prediction Rate  \n",
        "         Groups                                        \n",
-       "original Overall                             0.202867  \n",
-       "         0                                   0.080357  \n",
-       "         1                                   0.211082  \n",
-       "         2                                   0.073192  \n",
-       "         3                                   0.053571  \n",
-       "         4                                   0.219517  \n",
-       "         Maximum difference                  0.165945  \n",
-       "updated  Overall                             0.178215  \n",
-       "         0                                   0.178571  \n",
-       "         1                                   0.176781  \n",
-       "         2                                   0.134039  \n",
-       "         3                                   0.178571  \n",
-       "         4                                   0.183042  \n",
-       "         Maximum difference                  0.049003  "
+       "original Overall                             0.203030  \n",
+       "         0                                   0.047619  \n",
+       "         1                                   0.224932  \n",
+       "         2                                   0.096106  \n",
+       "         3                                   0.117021  \n",
+       "         4                                   0.217304  \n",
+       "         Maximum difference                  0.177313  \n",
+       "updated  Overall                             0.193612  \n",
+       "         0                                   0.166667  \n",
+       "         1                                   0.173442  \n",
+       "         2                                   0.159901  \n",
+       "         3                                   0.170213  \n",
+       "         4                                   0.198771  \n",
+       "         Maximum difference                  0.038870  "
       ]
      },
      "execution_count": 15,
@@ -1767,73 +1767,73 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Demographic Parity</th>\n",
-       "      <td>0.134287</td>\n",
-       "      <td>0.016033</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.650000</td>\n",
+       "      <td>0.170373</td>\n",
+       "      <td>0.063117</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.663866</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Disparate Impact</th>\n",
-       "      <td>0.706830</td>\n",
-       "      <td>0.981370</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.647479</td>\n",
+       "      <td>0.651723</td>\n",
+       "      <td>0.892211</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.663866</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Conditional Acceptance Rate</th>\n",
-       "      <td>0.197906</td>\n",
-       "      <td>0.008440</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.655462</td>\n",
+       "      <td>0.392026</td>\n",
+       "      <td>0.071445</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.673950</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Conditional Rejectance Rate</th>\n",
-       "      <td>0.046598</td>\n",
-       "      <td>0.016439</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.655462</td>\n",
+       "      <td>0.111950</td>\n",
+       "      <td>0.034648</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.674790</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.015688</td>\n",
-       "      <td>0.021157</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.654622</td>\n",
+       "      <td>0.022351</td>\n",
+       "      <td>0.025632</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.667647</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Recall</th>\n",
-       "      <td>0.141640</td>\n",
-       "      <td>0.058647</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.652101</td>\n",
+       "      <td>0.177895</td>\n",
+       "      <td>0.069213</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.671008</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Acceptance Rate</th>\n",
-       "      <td>0.087016</td>\n",
-       "      <td>0.076756</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.660504</td>\n",
+       "      <td>0.040866</td>\n",
+       "      <td>0.055503</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.670168</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Specificity</th>\n",
-       "      <td>0.091102</td>\n",
-       "      <td>0.053444</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.658824</td>\n",
+       "      <td>0.121835</td>\n",
+       "      <td>0.057111</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.674370</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Rejection Rate</th>\n",
-       "      <td>0.052209</td>\n",
-       "      <td>0.035084</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.660924</td>\n",
+       "      <td>0.050291</td>\n",
+       "      <td>0.038554</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.668487</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.185564</td>\n",
-       "      <td>0.055206</td>\n",
-       "      <td>0.657983</td>\n",
-       "      <td>0.655462</td>\n",
+       "      <td>0.360054</td>\n",
+       "      <td>0.052966</td>\n",
+       "      <td>0.671849</td>\n",
+       "      <td>0.673950</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -1841,52 +1841,52 @@
       ],
       "text/plain": [
        "                                                    Measure (original)  \\\n",
-       "Demographic Parity                                            0.134287   \n",
-       "Disparate Impact                                              0.706830   \n",
-       "Average Group Difference in Conditional Accepta...            0.197906   \n",
-       "Average Group Difference in Conditional Rejecta...            0.046598   \n",
-       "Average Group Difference in Accuracy                          0.015688   \n",
-       "Average Group Difference in Recall                            0.141640   \n",
-       "Average Group Difference in Acceptance Rate                   0.087016   \n",
-       "Average Group Difference in Specificity                       0.091102   \n",
-       "Average Group Difference in Rejection Rate                    0.052209   \n",
-       "Treatment Equality                                            0.185564   \n",
+       "Demographic Parity                                            0.170373   \n",
+       "Disparate Impact                                              0.651723   \n",
+       "Average Group Difference in Conditional Accepta...            0.392026   \n",
+       "Average Group Difference in Conditional Rejecta...            0.111950   \n",
+       "Average Group Difference in Accuracy                          0.022351   \n",
+       "Average Group Difference in Recall                            0.177895   \n",
+       "Average Group Difference in Acceptance Rate                   0.040866   \n",
+       "Average Group Difference in Specificity                       0.121835   \n",
+       "Average Group Difference in Rejection Rate                    0.050291   \n",
+       "Treatment Equality                                            0.360054   \n",
        "\n",
        "                                                    Measure (updated)  \\\n",
-       "Demographic Parity                                           0.016033   \n",
-       "Disparate Impact                                             0.981370   \n",
-       "Average Group Difference in Conditional Accepta...           0.008440   \n",
-       "Average Group Difference in Conditional Rejecta...           0.016439   \n",
-       "Average Group Difference in Accuracy                         0.021157   \n",
-       "Average Group Difference in Recall                           0.058647   \n",
-       "Average Group Difference in Acceptance Rate                  0.076756   \n",
-       "Average Group Difference in Specificity                      0.053444   \n",
-       "Average Group Difference in Rejection Rate                   0.035084   \n",
-       "Treatment Equality                                           0.055206   \n",
+       "Demographic Parity                                           0.063117   \n",
+       "Disparate Impact                                             0.892211   \n",
+       "Average Group Difference in Conditional Accepta...           0.071445   \n",
+       "Average Group Difference in Conditional Rejecta...           0.034648   \n",
+       "Average Group Difference in Accuracy                         0.025632   \n",
+       "Average Group Difference in Recall                           0.069213   \n",
+       "Average Group Difference in Acceptance Rate                  0.055503   \n",
+       "Average Group Difference in Specificity                      0.057111   \n",
+       "Average Group Difference in Rejection Rate                   0.038554   \n",
+       "Treatment Equality                                           0.052966   \n",
        "\n",
        "                                                    Accuracy (original)  \\\n",
-       "Demographic Parity                                             0.657983   \n",
-       "Disparate Impact                                               0.657983   \n",
-       "Average Group Difference in Conditional Accepta...             0.657983   \n",
-       "Average Group Difference in Conditional Rejecta...             0.657983   \n",
-       "Average Group Difference in Accuracy                           0.657983   \n",
-       "Average Group Difference in Recall                             0.657983   \n",
-       "Average Group Difference in Acceptance Rate                    0.657983   \n",
-       "Average Group Difference in Specificity                        0.657983   \n",
-       "Average Group Difference in Rejection Rate                     0.657983   \n",
-       "Treatment Equality                                             0.657983   \n",
+       "Demographic Parity                                             0.671849   \n",
+       "Disparate Impact                                               0.671849   \n",
+       "Average Group Difference in Conditional Accepta...             0.671849   \n",
+       "Average Group Difference in Conditional Rejecta...             0.671849   \n",
+       "Average Group Difference in Accuracy                           0.671849   \n",
+       "Average Group Difference in Recall                             0.671849   \n",
+       "Average Group Difference in Acceptance Rate                    0.671849   \n",
+       "Average Group Difference in Specificity                        0.671849   \n",
+       "Average Group Difference in Rejection Rate                     0.671849   \n",
+       "Treatment Equality                                             0.671849   \n",
        "\n",
        "                                                    Accuracy (updated)  \n",
-       "Demographic Parity                                            0.650000  \n",
-       "Disparate Impact                                              0.647479  \n",
-       "Average Group Difference in Conditional Accepta...            0.655462  \n",
-       "Average Group Difference in Conditional Rejecta...            0.655462  \n",
-       "Average Group Difference in Accuracy                          0.654622  \n",
-       "Average Group Difference in Recall                            0.652101  \n",
-       "Average Group Difference in Acceptance Rate                   0.660504  \n",
-       "Average Group Difference in Specificity                       0.658824  \n",
-       "Average Group Difference in Rejection Rate                    0.660924  \n",
-       "Treatment Equality                                            0.655462  "
+       "Demographic Parity                                            0.663866  \n",
+       "Disparate Impact                                              0.663866  \n",
+       "Average Group Difference in Conditional Accepta...            0.673950  \n",
+       "Average Group Difference in Conditional Rejecta...            0.674790  \n",
+       "Average Group Difference in Accuracy                          0.667647  \n",
+       "Average Group Difference in Recall                            0.671008  \n",
+       "Average Group Difference in Acceptance Rate                   0.670168  \n",
+       "Average Group Difference in Specificity                       0.674370  \n",
+       "Average Group Difference in Rejection Rate                    0.668487  \n",
+       "Treatment Equality                                            0.673950  "
       ]
      },
      "execution_count": 21,
@@ -1948,73 +1948,73 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Demographic Parity</th>\n",
-       "      <td>0.153068</td>\n",
-       "      <td>0.102293</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.621429</td>\n",
+       "      <td>0.142173</td>\n",
+       "      <td>0.107317</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.624370</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Disparate Impact</th>\n",
-       "      <td>0.697450</td>\n",
-       "      <td>0.686713</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.611765</td>\n",
+       "      <td>0.682094</td>\n",
+       "      <td>0.731433</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.618487</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Conditional Acceptance Rate</th>\n",
-       "      <td>0.266809</td>\n",
-       "      <td>0.155159</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.647059</td>\n",
+       "      <td>0.287004</td>\n",
+       "      <td>0.163108</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.643697</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Conditional Rejectance Rate</th>\n",
-       "      <td>0.098374</td>\n",
-       "      <td>0.038179</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.661765</td>\n",
+       "      <td>0.068234</td>\n",
+       "      <td>0.063120</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.649580</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Accuracy</th>\n",
-       "      <td>0.005545</td>\n",
-       "      <td>0.019626</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.661345</td>\n",
+       "      <td>0.028167</td>\n",
+       "      <td>0.026010</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.650000</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Recall</th>\n",
-       "      <td>0.157070</td>\n",
-       "      <td>0.118976</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.628571</td>\n",
+       "      <td>0.143907</td>\n",
+       "      <td>0.130195</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.628992</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Acceptance Rate</th>\n",
-       "      <td>0.078017</td>\n",
-       "      <td>0.084274</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.657983</td>\n",
+       "      <td>0.059639</td>\n",
+       "      <td>0.061339</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.660924</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Specificity</th>\n",
-       "      <td>0.107762</td>\n",
-       "      <td>0.089286</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.649160</td>\n",
+       "      <td>0.103289</td>\n",
+       "      <td>0.089840</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.645378</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Rejection Rate</th>\n",
-       "      <td>0.043247</td>\n",
-       "      <td>0.041363</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.660504</td>\n",
+       "      <td>0.073389</td>\n",
+       "      <td>0.062303</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.655882</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Treatment Equality</th>\n",
-       "      <td>0.318437</td>\n",
-       "      <td>0.143180</td>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.646639</td>\n",
+       "      <td>0.237167</td>\n",
+       "      <td>0.160742</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.646218</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -2022,52 +2022,52 @@
       ],
       "text/plain": [
        "                                                    Measure (original)  \\\n",
-       "Demographic Parity                                            0.153068   \n",
-       "Disparate Impact                                              0.697450   \n",
-       "Average Group Difference in Conditional Accepta...            0.266809   \n",
-       "Average Group Difference in Conditional Rejecta...            0.098374   \n",
-       "Average Group Difference in Accuracy                          0.005545   \n",
-       "Average Group Difference in Recall                            0.157070   \n",
-       "Average Group Difference in Acceptance Rate                   0.078017   \n",
-       "Average Group Difference in Specificity                       0.107762   \n",
-       "Average Group Difference in Rejection Rate                    0.043247   \n",
-       "Treatment Equality                                            0.318437   \n",
+       "Demographic Parity                                            0.142173   \n",
+       "Disparate Impact                                              0.682094   \n",
+       "Average Group Difference in Conditional Accepta...            0.287004   \n",
+       "Average Group Difference in Conditional Rejecta...            0.068234   \n",
+       "Average Group Difference in Accuracy                          0.028167   \n",
+       "Average Group Difference in Recall                            0.143907   \n",
+       "Average Group Difference in Acceptance Rate                   0.059639   \n",
+       "Average Group Difference in Specificity                       0.103289   \n",
+       "Average Group Difference in Rejection Rate                    0.073389   \n",
+       "Treatment Equality                                            0.237167   \n",
        "\n",
        "                                                    Measure (updated)  \\\n",
-       "Demographic Parity                                           0.102293   \n",
-       "Disparate Impact                                             0.686713   \n",
-       "Average Group Difference in Conditional Accepta...           0.155159   \n",
-       "Average Group Difference in Conditional Rejecta...           0.038179   \n",
-       "Average Group Difference in Accuracy                         0.019626   \n",
-       "Average Group Difference in Recall                           0.118976   \n",
-       "Average Group Difference in Acceptance Rate                  0.084274   \n",
-       "Average Group Difference in Specificity                      0.089286   \n",
-       "Average Group Difference in Rejection Rate                   0.041363   \n",
-       "Treatment Equality                                           0.143180   \n",
+       "Demographic Parity                                           0.107317   \n",
+       "Disparate Impact                                             0.731433   \n",
+       "Average Group Difference in Conditional Accepta...           0.163108   \n",
+       "Average Group Difference in Conditional Rejecta...           0.063120   \n",
+       "Average Group Difference in Accuracy                         0.026010   \n",
+       "Average Group Difference in Recall                           0.130195   \n",
+       "Average Group Difference in Acceptance Rate                  0.061339   \n",
+       "Average Group Difference in Specificity                      0.089840   \n",
+       "Average Group Difference in Rejection Rate                   0.062303   \n",
+       "Treatment Equality                                           0.160742   \n",
        "\n",
        "                                                    Accuracy (original)  \\\n",
-       "Demographic Parity                                             0.663866   \n",
-       "Disparate Impact                                               0.663866   \n",
-       "Average Group Difference in Conditional Accepta...             0.663866   \n",
-       "Average Group Difference in Conditional Rejecta...             0.663866   \n",
-       "Average Group Difference in Accuracy                           0.663866   \n",
-       "Average Group Difference in Recall                             0.663866   \n",
-       "Average Group Difference in Acceptance Rate                    0.663866   \n",
-       "Average Group Difference in Specificity                        0.663866   \n",
-       "Average Group Difference in Rejection Rate                     0.663866   \n",
-       "Treatment Equality                                             0.663866   \n",
+       "Demographic Parity                                             0.653361   \n",
+       "Disparate Impact                                               0.653361   \n",
+       "Average Group Difference in Conditional Accepta...             0.653361   \n",
+       "Average Group Difference in Conditional Rejecta...             0.653361   \n",
+       "Average Group Difference in Accuracy                           0.653361   \n",
+       "Average Group Difference in Recall                             0.653361   \n",
+       "Average Group Difference in Acceptance Rate                    0.653361   \n",
+       "Average Group Difference in Specificity                        0.653361   \n",
+       "Average Group Difference in Rejection Rate                     0.653361   \n",
+       "Treatment Equality                                             0.653361   \n",
        "\n",
        "                                                    Accuracy (updated)  \n",
-       "Demographic Parity                                            0.621429  \n",
-       "Disparate Impact                                              0.611765  \n",
-       "Average Group Difference in Conditional Accepta...            0.647059  \n",
-       "Average Group Difference in Conditional Rejecta...            0.661765  \n",
-       "Average Group Difference in Accuracy                          0.661345  \n",
-       "Average Group Difference in Recall                            0.628571  \n",
-       "Average Group Difference in Acceptance Rate                   0.657983  \n",
-       "Average Group Difference in Specificity                       0.649160  \n",
-       "Average Group Difference in Rejection Rate                    0.660504  \n",
-       "Treatment Equality                                            0.646639  "
+       "Demographic Parity                                            0.624370  \n",
+       "Disparate Impact                                              0.618487  \n",
+       "Average Group Difference in Conditional Accepta...            0.643697  \n",
+       "Average Group Difference in Conditional Rejecta...            0.649580  \n",
+       "Average Group Difference in Accuracy                          0.650000  \n",
+       "Average Group Difference in Recall                            0.628992  \n",
+       "Average Group Difference in Acceptance Rate                   0.660924  \n",
+       "Average Group Difference in Specificity                       0.645378  \n",
+       "Average Group Difference in Rejection Rate                    0.655882  \n",
+       "Treatment Equality                                            0.646218  "
       ]
      },
      "execution_count": 22,
@@ -2150,144 +2150,144 @@
        "    <tr>\n",
        "      <th rowspan=\"5\" valign=\"top\">original</th>\n",
        "      <th>Overall</th>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.655245</td>\n",
-       "      <td>0.603568</td>\n",
-       "      <td>0.315750</td>\n",
-       "      <td>0.644444</td>\n",
-       "      <td>0.567568</td>\n",
-       "      <td>0.705639</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.642428</td>\n",
+       "      <td>0.580153</td>\n",
+       "      <td>0.292805</td>\n",
+       "      <td>0.639013</td>\n",
+       "      <td>0.531221</td>\n",
+       "      <td>0.700889</td>\n",
        "      <td>1073.0</td>\n",
        "      <td>1307.0</td>\n",
        "      <td>0.450840</td>\n",
-       "      <td>0.397059</td>\n",
+       "      <td>0.374790</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>African-American</th>\n",
-       "      <td>0.661198</td>\n",
-       "      <td>0.661083</td>\n",
-       "      <td>0.668805</td>\n",
-       "      <td>0.322081</td>\n",
-       "      <td>0.672581</td>\n",
-       "      <td>0.665072</td>\n",
-       "      <td>0.704855</td>\n",
+       "      <td>0.641509</td>\n",
+       "      <td>0.642416</td>\n",
+       "      <td>0.636741</td>\n",
+       "      <td>0.285145</td>\n",
+       "      <td>0.664931</td>\n",
+       "      <td>0.610845</td>\n",
+       "      <td>0.701756</td>\n",
        "      <td>627.0</td>\n",
        "      <td>592.0</td>\n",
        "      <td>0.514356</td>\n",
-       "      <td>0.508614</td>\n",
+       "      <td>0.472518</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Caucasian</th>\n",
-       "      <td>0.665432</td>\n",
-       "      <td>0.624102</td>\n",
-       "      <td>0.502752</td>\n",
-       "      <td>0.270386</td>\n",
-       "      <td>0.606195</td>\n",
-       "      <td>0.429467</td>\n",
-       "      <td>0.683915</td>\n",
+       "      <td>0.658025</td>\n",
+       "      <td>0.611952</td>\n",
+       "      <td>0.476371</td>\n",
+       "      <td>0.249640</td>\n",
+       "      <td>0.600000</td>\n",
+       "      <td>0.394984</td>\n",
+       "      <td>0.673311</td>\n",
        "      <td>319.0</td>\n",
        "      <td>491.0</td>\n",
        "      <td>0.393827</td>\n",
-       "      <td>0.279012</td>\n",
+       "      <td>0.259259</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Other</th>\n",
-       "      <td>0.669516</td>\n",
-       "      <td>0.618321</td>\n",
-       "      <td>0.486726</td>\n",
-       "      <td>0.252697</td>\n",
-       "      <td>0.555556</td>\n",
-       "      <td>0.433071</td>\n",
-       "      <td>0.703881</td>\n",
+       "      <td>0.683761</td>\n",
+       "      <td>0.639711</td>\n",
+       "      <td>0.523605</td>\n",
+       "      <td>0.292449</td>\n",
+       "      <td>0.575472</td>\n",
+       "      <td>0.480315</td>\n",
+       "      <td>0.688607</td>\n",
        "      <td>127.0</td>\n",
        "      <td>224.0</td>\n",
        "      <td>0.361823</td>\n",
-       "      <td>0.282051</td>\n",
+       "      <td>0.301994</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Maximum difference</th>\n",
-       "      <td>0.008318</td>\n",
-       "      <td>0.042762</td>\n",
-       "      <td>0.182079</td>\n",
-       "      <td>0.069384</td>\n",
-       "      <td>0.117025</td>\n",
-       "      <td>0.235605</td>\n",
-       "      <td>0.020940</td>\n",
+       "      <td>0.042251</td>\n",
+       "      <td>0.030463</td>\n",
+       "      <td>0.160371</td>\n",
+       "      <td>0.042809</td>\n",
+       "      <td>0.089459</td>\n",
+       "      <td>0.215861</td>\n",
+       "      <td>0.028445</td>\n",
        "      <td>500.0</td>\n",
        "      <td>368.0</td>\n",
        "      <td>0.152533</td>\n",
-       "      <td>0.229601</td>\n",
+       "      <td>0.213259</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th rowspan=\"5\" valign=\"top\">updated</th>\n",
        "      <th>Overall</th>\n",
-       "      <td>0.646639</td>\n",
-       "      <td>0.629716</td>\n",
-       "      <td>0.538673</td>\n",
-       "      <td>0.277867</td>\n",
-       "      <td>0.654667</td>\n",
-       "      <td>0.457596</td>\n",
-       "      <td>0.692327</td>\n",
+       "      <td>0.646218</td>\n",
+       "      <td>0.632754</td>\n",
+       "      <td>0.558237</td>\n",
+       "      <td>0.276979</td>\n",
+       "      <td>0.638655</td>\n",
+       "      <td>0.495806</td>\n",
+       "      <td>0.688616</td>\n",
        "      <td>1073.0</td>\n",
        "      <td>1307.0</td>\n",
        "      <td>0.450840</td>\n",
-       "      <td>0.315126</td>\n",
+       "      <td>0.350000</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>African-American</th>\n",
-       "      <td>0.639869</td>\n",
-       "      <td>0.642660</td>\n",
-       "      <td>0.609083</td>\n",
-       "      <td>0.290279</td>\n",
-       "      <td>0.689516</td>\n",
-       "      <td>0.545455</td>\n",
-       "      <td>0.690321</td>\n",
+       "      <td>0.634947</td>\n",
+       "      <td>0.636885</td>\n",
+       "      <td>0.616048</td>\n",
+       "      <td>0.275896</td>\n",
+       "      <td>0.671053</td>\n",
+       "      <td>0.569378</td>\n",
+       "      <td>0.693082</td>\n",
        "      <td>627.0</td>\n",
        "      <td>592.0</td>\n",
        "      <td>0.514356</td>\n",
-       "      <td>0.406891</td>\n",
+       "      <td>0.436423</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Caucasian</th>\n",
-       "      <td>0.644444</td>\n",
-       "      <td>0.590868</td>\n",
-       "      <td>0.428571</td>\n",
-       "      <td>0.211519</td>\n",
-       "      <td>0.583784</td>\n",
-       "      <td>0.338558</td>\n",
-       "      <td>0.667708</td>\n",
+       "      <td>0.654321</td>\n",
+       "      <td>0.605603</td>\n",
+       "      <td>0.461538</td>\n",
+       "      <td>0.238911</td>\n",
+       "      <td>0.597015</td>\n",
+       "      <td>0.376176</td>\n",
+       "      <td>0.655683</td>\n",
        "      <td>319.0</td>\n",
        "      <td>491.0</td>\n",
        "      <td>0.393827</td>\n",
-       "      <td>0.228395</td>\n",
+       "      <td>0.248148</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Other</th>\n",
-       "      <td>0.675214</td>\n",
-       "      <td>0.598917</td>\n",
-       "      <td>0.418367</td>\n",
-       "      <td>0.239210</td>\n",
-       "      <td>0.594203</td>\n",
-       "      <td>0.322835</td>\n",
-       "      <td>0.705884</td>\n",
+       "      <td>0.666667</td>\n",
+       "      <td>0.616089</td>\n",
+       "      <td>0.484581</td>\n",
+       "      <td>0.247178</td>\n",
+       "      <td>0.550000</td>\n",
+       "      <td>0.433071</td>\n",
+       "      <td>0.673703</td>\n",
        "      <td>127.0</td>\n",
        "      <td>224.0</td>\n",
        "      <td>0.361823</td>\n",
-       "      <td>0.196581</td>\n",
+       "      <td>0.284900</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Maximum difference</th>\n",
-       "      <td>0.035345</td>\n",
-       "      <td>0.051792</td>\n",
-       "      <td>0.190715</td>\n",
-       "      <td>0.078760</td>\n",
-       "      <td>0.105732</td>\n",
-       "      <td>0.222620</td>\n",
-       "      <td>0.038176</td>\n",
+       "      <td>0.031720</td>\n",
+       "      <td>0.031282</td>\n",
+       "      <td>0.154510</td>\n",
+       "      <td>0.036986</td>\n",
+       "      <td>0.121053</td>\n",
+       "      <td>0.193202</td>\n",
+       "      <td>0.037399</td>\n",
        "      <td>500.0</td>\n",
        "      <td>368.0</td>\n",
        "      <td>0.152533</td>\n",
-       "      <td>0.210310</td>\n",
+       "      <td>0.188275</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -2296,29 +2296,29 @@
       "text/plain": [
        "                             Accuracy  Balanced Accuracy  F1 score       MCC  \\\n",
        "         Groups                                                                \n",
-       "original Overall             0.663866           0.655245  0.603568  0.315750   \n",
-       "         African-American    0.661198           0.661083  0.668805  0.322081   \n",
-       "         Caucasian           0.665432           0.624102  0.502752  0.270386   \n",
-       "         Other               0.669516           0.618321  0.486726  0.252697   \n",
-       "         Maximum difference  0.008318           0.042762  0.182079  0.069384   \n",
-       "updated  Overall             0.646639           0.629716  0.538673  0.277867   \n",
-       "         African-American    0.639869           0.642660  0.609083  0.290279   \n",
-       "         Caucasian           0.644444           0.590868  0.428571  0.211519   \n",
-       "         Other               0.675214           0.598917  0.418367  0.239210   \n",
-       "         Maximum difference  0.035345           0.051792  0.190715  0.078760   \n",
+       "original Overall             0.653361           0.642428  0.580153  0.292805   \n",
+       "         African-American    0.641509           0.642416  0.636741  0.285145   \n",
+       "         Caucasian           0.658025           0.611952  0.476371  0.249640   \n",
+       "         Other               0.683761           0.639711  0.523605  0.292449   \n",
+       "         Maximum difference  0.042251           0.030463  0.160371  0.042809   \n",
+       "updated  Overall             0.646218           0.632754  0.558237  0.276979   \n",
+       "         African-American    0.634947           0.636885  0.616048  0.275896   \n",
+       "         Caucasian           0.654321           0.605603  0.461538  0.238911   \n",
+       "         Other               0.666667           0.616089  0.484581  0.247178   \n",
+       "         Maximum difference  0.031720           0.031282  0.154510  0.036986   \n",
        "\n",
        "                             Precision    Recall   ROC AUC  Positive Count  \\\n",
        "         Groups                                                              \n",
-       "original Overall              0.644444  0.567568  0.705639          1073.0   \n",
-       "         African-American     0.672581  0.665072  0.704855           627.0   \n",
-       "         Caucasian            0.606195  0.429467  0.683915           319.0   \n",
-       "         Other                0.555556  0.433071  0.703881           127.0   \n",
-       "         Maximum difference   0.117025  0.235605  0.020940           500.0   \n",
-       "updated  Overall              0.654667  0.457596  0.692327          1073.0   \n",
-       "         African-American     0.689516  0.545455  0.690321           627.0   \n",
-       "         Caucasian            0.583784  0.338558  0.667708           319.0   \n",
-       "         Other                0.594203  0.322835  0.705884           127.0   \n",
-       "         Maximum difference   0.105732  0.222620  0.038176           500.0   \n",
+       "original Overall              0.639013  0.531221  0.700889          1073.0   \n",
+       "         African-American     0.664931  0.610845  0.701756           627.0   \n",
+       "         Caucasian            0.600000  0.394984  0.673311           319.0   \n",
+       "         Other                0.575472  0.480315  0.688607           127.0   \n",
+       "         Maximum difference   0.089459  0.215861  0.028445           500.0   \n",
+       "updated  Overall              0.638655  0.495806  0.688616          1073.0   \n",
+       "         African-American     0.671053  0.569378  0.693082           627.0   \n",
+       "         Caucasian            0.597015  0.376176  0.655683           319.0   \n",
+       "         Other                0.550000  0.433071  0.673703           127.0   \n",
+       "         Maximum difference   0.121053  0.193202  0.037399           500.0   \n",
        "\n",
        "                             Negative Count  Positive Label Rate  \\\n",
        "         Groups                                                    \n",
@@ -2335,16 +2335,16 @@
        "\n",
        "                             Positive Prediction Rate  \n",
        "         Groups                                        \n",
-       "original Overall                             0.397059  \n",
-       "         African-American                    0.508614  \n",
-       "         Caucasian                           0.279012  \n",
-       "         Other                               0.282051  \n",
-       "         Maximum difference                  0.229601  \n",
-       "updated  Overall                             0.315126  \n",
-       "         African-American                    0.406891  \n",
-       "         Caucasian                           0.228395  \n",
-       "         Other                               0.196581  \n",
-       "         Maximum difference                  0.210310  "
+       "original Overall                             0.374790  \n",
+       "         African-American                    0.472518  \n",
+       "         Caucasian                           0.259259  \n",
+       "         Other                               0.301994  \n",
+       "         Maximum difference                  0.213259  \n",
+       "updated  Overall                             0.350000  \n",
+       "         African-American                    0.436423  \n",
+       "         Caucasian                           0.248148  \n",
+       "         Other                               0.284900  \n",
+       "         Maximum difference                  0.188275  "
       ]
      },
      "execution_count": 23,
@@ -2414,13 +2414,13 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.736455</td>\n",
-       "      <td>0.681856</td>\n",
+       "      <td>0.738975</td>\n",
+       "      <td>0.693406</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Conditional Positive Prediction Rate</th>\n",
-       "      <td>0.136081</td>\n",
-       "      <td>0.019586</td>\n",
+       "      <td>0.137529</td>\n",
+       "      <td>0.019887</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -2428,8 +2428,8 @@
       ],
       "text/plain": [
        "                                                    original   updated\n",
-       "Accuracy                                            0.736455  0.681856\n",
-       "Average Group Difference in Conditional Positiv...  0.136081  0.019586"
+       "Accuracy                                            0.738975  0.693406\n",
+       "Average Group Difference in Conditional Positiv...  0.137529  0.019887"
       ]
      },
      "execution_count": 25,
@@ -2481,13 +2481,13 @@
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>Accuracy</th>\n",
-       "      <td>0.663866</td>\n",
-       "      <td>0.619328</td>\n",
+       "      <td>0.653361</td>\n",
+       "      <td>0.626471</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Average Group Difference in Conditional Positive Prediction Rate</th>\n",
-       "      <td>0.151770</td>\n",
-       "      <td>0.111426</td>\n",
+       "      <td>0.136923</td>\n",
+       "      <td>0.100760</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
@@ -2495,8 +2495,8 @@
       ],
       "text/plain": [
        "                                                    original   updated\n",
-       "Accuracy                                            0.663866  0.619328\n",
-       "Average Group Difference in Conditional Positiv...  0.151770  0.111426"
+       "Accuracy                                            0.653361  0.626471\n",
+       "Average Group Difference in Conditional Positiv...  0.136923  0.100760"
       ]
      },
      "execution_count": 26,
@@ -2525,7 +2525,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.1.-1"
+   "version": "3.10.13"
   }
  },
  "nbformat": 4,