[14w] -5 classification plots.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "a2cf475e",
   "metadata": {},
   "source": [
    "# library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "4c6ae050",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-06-18 14:34:26.070783: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.10.1\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from IPython.core.interactiveshell import InteractiveShell\n",
    "InteractiveShell.ast_node_interactivity = 'all'\n",
    "\n",
    "import numpy as np\n",
    "from tqdm import tqdm\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "import tensorflow as tf\n",
    "import random"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f0f0b1a1",
   "metadata": {},
   "source": [
    "## mk_cd 함수"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "4ad1d5d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "## 함수\n",
    "def mk_cd(x):\n",
    "    if type(x) == str:\n",
    "        if x.isalpha():\n",
    "            return x\n",
    "        else:\n",
    "            if len(x) == 6:\n",
    "                return x\n",
    "            else:\n",
    "                return \"0\"*(6-len(x))+x\n",
    "    elif type(x) == int:\n",
    "        x = str(x)\n",
    "        return \"0\"*(6-len(x))+x"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "72637196",
   "metadata": {},
   "source": [
    "# 슬라이싱 준비"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "4bace3aa",
   "metadata": {},
   "outputs": [],
   "source": [
    "len_a2012 = 284412\n",
    "len_a2013 = 242826\n",
    "len_a2014 = 242448\n",
    "len_a2015 = 314361\n",
    "len_a2016 = 263593\n",
    "len_a2017 = 216591\n",
    "len_a2018 = 307395\n",
    "len_a2019 = 278610\n",
    "len_a2020 = 371006\n",
    "len_a2021 = 321808"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "1294cf86",
   "metadata": {},
   "outputs": [],
   "source": [
    "len1215 = len_a2012+len_a2013+len_a2014+len_a2015 # y:2015\n",
    "len1316 = len1215+len_a2016 # y:2016\n",
    "len1417 = len1316+len_a2017 # y:2017\n",
    "len1518 = len1417+len_a2018 # y:2018\n",
    "len1619 = len1518+len_a2019 # y:2019\n",
    "len1720 = len1619+len_a2020 # y:2020\n",
    "len1821 = len1720+len_a2021 # y:2021"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2dfdbd29",
   "metadata": {},
   "source": [
    "## [0], [1]은 train 슬라이싱 인덱스, [2]는 test인덱스로 구성"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "eae1dcdf",
   "metadata": {},
   "outputs": [],
   "source": [
    "slicing_dic = {'1215':[0, len1215, len_a2015],\n",
    "               '1316':[len_a2012, len1316, len_a2016],\n",
    "               '1417':[len_a2012+len_a2013, len1417, len_a2017],\n",
    "               '1518':[len_a2012+len_a2013+len_a2014, len1518, len_a2018],\n",
    "               '1619':[len_a2012+len_a2013+len_a2014+len_a2015, len1619, len_a2019],\n",
    "               '1720':[len_a2012+len_a2013+len_a2014+len_a2015+len_a2016,len1720, len_a2020],\n",
    "               '1821':[len_a2012+len_a2013+len_a2014+len_a2015+len_a2016+len_a2017,len1821, len_a2021]}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "709ddf1b",
   "metadata": {},
   "source": [
    "## 슬라이싱 테스트"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "4b0edf41",
   "metadata": {},
   "outputs": [],
   "source": [
    "# d1221[:len1215]\n",
    "# d1221[len_a2012:len1316]\n",
    "# d1221[len_a2012+len_a2013:len1417]\n",
    "# d1221[len_a2012+len_a2013+len_a2014:len1518]\n",
    "# d1221[len_a2012+len_a2013+len_a2014+len_a2015:len1619]\n",
    "# d1221[len_a2012+len_a2013+len_a2014+len_a2015+len_a2016:len1720]\n",
    "# d1221[len_a2012+len_a2013+len_a2014+len_a2015+len_a2016+len_a2017:len1821]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a81d8b16",
   "metadata": {},
   "source": [
    "# data load"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "bcc7c845",
   "metadata": {},
   "outputs": [],
   "source": [
    "d1221 = pd.read_feather('d1221.ftr', columns = None, use_threads = True) # threads 사용을 꺼도 난리 (그래도 비교적 덜 난리)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17ed0484",
   "metadata": {},
   "source": [
    "# category to object"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ae7ccf2e",
   "metadata": {},
   "outputs": [],
   "source": [
    "d1221['code'] = list(map(mk_cd, d1221['code'])) # mk_cd함수 적용시켜야 ftr만들 때 에러 안남."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "08aa8296",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 2843050 entries, 0 to 2843049\n",
      "Columns: 823 entries, date to NextChange\n",
      "dtypes: float64(821), int64(1), object(1)\n",
      "memory usage: 17.4+ GB\n"
     ]
    }
   ],
   "source": [
    "d1221.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6c42a34",
   "metadata": {},
   "source": [
    "# modeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "98b8c4d1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# NextChange\n",
    "d1221['NextChange'] = np.where(d1221['NextChange']>0.05, 1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "a1fc31fd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1297139"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "248772"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "0.08750180264152935"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "2843050"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnt0 = d1221.shape[0] - d1221['NextChange'].sum()\n",
    "cnt1 = d1221['NextChange'].sum()\n",
    "cnt1\n",
    "cnt0-cnt1\n",
    "(cnt0-cnt1) / d1221.shape[0]\n",
    "d1221.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "8c1dfdec",
   "metadata": {},
   "outputs": [],
   "source": [
    "from datetime import datetime\n",
    "# 1. Logistic regression\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "# 2. Decision tree\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "# 3. Support vector machine\n",
    "from sklearn.svm import SVC\n",
    "# 4. Gaussian naive bayes\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "# 5. K nearest neighbor\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "# 6. Random forest\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "# 7. Gradient boosing\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "# 8. Neural network\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "# import lightgbm as lgb\n",
    "from xgboost import XGBClassifier # decision tree 앙상블 모델, 부스팅\n",
    "# [light gbm 장점, 하이퍼파라미터 튜닝 방안](https://ariz1623.tistory.com/209)\n",
    "# LightGBM도 XGBoost와 동일하게 조기 중단 수행 가능.\n",
    "# [XGBoost와 LightGBM 하이퍼파라미터 튜닝 가이드](https://psystat.tistory.com/131)\n",
    "\n",
    "\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "from sklearn.dummy import DummyClassifier\n",
    "\n",
    "# 모델 평가 지표 scoring metrics\n",
    "from sklearn.model_selection import cross_val_score # model 검증\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import f1_score\n",
    "from sklearn.metrics import jaccard_score\n",
    "\n",
    "import ast # convert string to function\n",
    "\n",
    "# [scikit-learn classifier metrics](https://scikit-learn.org/stable/modules/model_evaluation.html)\n",
    "# [classification metrics 1](https://hong-yp-ml-records.tistory.com/29)\n",
    "# [classification metrics 2](https://sw-data.tistory.com/23) -> regression metrics 참고\n",
    "\n",
    "results = []\n",
    "def my_classifier():\n",
    "    \n",
    "    # X, y\n",
    "    X = d1221.iloc[:,2:-1] # 1221 데이터\n",
    "    y = d1221.iloc[:,-1]\n",
    "    \n",
    "    # seed 고정\n",
    "    user_seed = 0\n",
    "    random.seed(user_seed) # seed 고정\n",
    "    \n",
    "    # 모델\n",
    "    # (같은 줄에 써놓은 애들끼리 하이퍼파리미터 구성 비슷하거나 같음)\n",
    "#     Logistic_Regression = LogisticRegression(n_jobs=64,\n",
    "#                                              random_state=user_seed,\n",
    "#                                              max_iter=10) # , verbose = 1\n",
    "    # [모델훈련](https://taek98.tistory.com/15)\n",
    "\n",
    "    DecisionTree = DecisionTreeClassifier(max_depth = 10,\n",
    "                                          random_state=user_seed,\n",
    "                                          max_leaf_nodes=4)\n",
    "    \n",
    "    RandomForest = RandomForestClassifier(n_estimators=10,\n",
    "                                          max_depth=10,\n",
    "                                          random_state=user_seed,\n",
    "                                          max_leaf_nodes=4,\n",
    "                                          n_jobs=64) # , verbose=1\n",
    "    \n",
    "    GradientBoosting = GradientBoostingClassifier(n_estimators=10,\n",
    "                                                  max_depth = 10,\n",
    "                                                  random_state = user_seed,\n",
    "                                                  max_leaf_nodes=4,\n",
    "                                                  learning_rate=0.05) # , verbose = 1\n",
    "    \n",
    "#     SVM = SVC(max_iter=1000,\n",
    "#               random_state=user_seed) # , verbose=True\n",
    "    \n",
    "#     Gaussian_NB = GaussianNB()\n",
    "    KNeighbors = KNeighborsClassifier(n_jobs = 64)\n",
    "    MLP = MLPClassifier(max_iter=1000,\n",
    "                        batch_size = 10000,\n",
    "                        shuffle=True,\n",
    "                        random_state = user_seed,\n",
    "                        early_stopping=True) # early_stopping 있어서 max_iter 키움., verbose=True\n",
    "\n",
    "#     XGBoost = XGBClassifier(n_jobs=64,\n",
    "#                             max_depth=10,\n",
    "#                             n_estimators=10,\n",
    "#                             learning_rate=0.05,\n",
    "#                             random_state = user_seed) # early_stopping_rounds = 50,\n",
    "    # [XGBoost Classifier hyper params](https://xgboost.readthedocs.io/en/stable/python/python_api.html)\n",
    "#     LGBM = lgb.LGBMClassifier(num_leaves=15,\n",
    "#                               learning_rate=0.05,\n",
    "#                               n_estimators=10,\n",
    "#                               max_depth = 10,\n",
    "#                               random_state = user_seed) # early_stopping_rounds, best_iteration_,\n",
    "    # max_depth : 과적합 방지를 위해 깊이 크기 제한\n",
    "    # n_estimators : 너무 크면 과적합, 성능저하\n",
    "    # learning_rate 작게 하면서 n_estimators를 크게 하는 것은 부스팅 계열 튜닝에서 가장 기본적인 튜닝 방안이므로 이를 적용하는 것도 좋다.\n",
    "    # [LGBM](https://lightgbm.readthedocs.io/en/latest/pythonapi/lightgbm.LGBMClassifier.html)\n",
    "    \n",
    "    # [gbm grid CV, best_iteration_ 참고](https://github.com/microsoft/LightGBM/blob/master/examples/python-guide/sklearn_example.py)\n",
    "    # [how to save model to best iteration if early stopping does not happen?](https://github.com/Microsoft/LightGBM/issues/1683)\n",
    "    \n",
    "    dummy = DummyClassifier(strategy = 'stratified', random_state = user_seed) # strategy 옵션: 'stratified', 'uniform', 'most_frequent'\n",
    "    \n",
    "    my_model_list = ['MLP',\n",
    "                     'DecisionTree', 'RandomForest', 'GradientBoosting','KNeighbors'] # 오래 걸리는걸 앞으로 뺌\n",
    "    # 'LGBM', 서버 자꾸 터져서 뺍니다\n",
    "    # 'XGBoost', 'dummy', 'Logistic_Regression', 'SVM', 'Gaussian_NB', 끝나서 뺍니다.\n",
    "    \n",
    "    i = 0\n",
    "    \n",
    "    ######################################################################### fitting\n",
    "    for model_nm in tqdm(my_model_list):\n",
    "        print(f'<<<---{model_nm} start--->>>')\n",
    "        print(datetime.now().strftime('%H:%M:%S'))\n",
    "        cntcnt = 0\n",
    "        # cv # 년도 나눠지는 것에 따라서 개수 정함.\n",
    "        for slicer in slicing_dic:\n",
    "            print(f'<<<---{slicer} start--->>>')\n",
    "            print(datetime.now().strftime('%H:%M:%S'))\n",
    "            tmpX = X[slicing_dic[slicer][0]:slicing_dic[slicer][1]].reset_index(drop = True) # X를 연속 4년치 꺼내서 reset_index(drop = True)\n",
    "            tmpy = y[slicing_dic[slicer][0]:slicing_dic[slicer][1]].reset_index(drop = True) # y를 연속 4년치 꺼내서 reset_index(drop = True)\n",
    "#             tmp_all = d1221[slicing_dic[slicer][0]:slicing_dic[slicer][1]].reset_index(drop = True) # 전체데이터\n",
    "\n",
    "            X_train = tmpX[:-slicing_dic[slicer][2]] # 기준년도 앞\n",
    "            X_test = tmpX[-slicing_dic[slicer][2]:] # 기준년도 이하\n",
    "            y_train = tmpy[:-slicing_dic[slicer][2]] # 기준년도 앞\n",
    "            y_test = tmpy[-slicing_dic[slicer][2]:] # 기준년도 이하\n",
    "            \n",
    "            # train끼리 idx 같아야 하므로\n",
    "            len_train = slicing_dic[slicer][1]-(slicing_dic[slicer][2]+slicing_dic[slicer][0]) # 혹은 tmp_train.shape[0]\n",
    "            len(list(range(len_train)))\n",
    "            train_idx = list(range(len_train))\n",
    "            random.shuffle(train_idx) # 자동으로 덮어쓰기\n",
    "            \n",
    "            # test끼리 idx 같아야 하므로\n",
    "            len_train_to_test = slicing_dic[slicer][1]-slicing_dic[slicer][0]\n",
    "            test_idx = list(range(len_train,len_train_to_test)) # +1 안해줘도 되는 거 위 cell에서 확인\n",
    "            len(test_idx)\n",
    "            random.shuffle(test_idx) # 자동으로 덮어쓰기\n",
    "            \n",
    "            X_train = X_train.loc[train_idx]\n",
    "            y_train = y_train.loc[train_idx]\n",
    "            X_test = X_test.loc[test_idx]\n",
    "            y_test = y_test.loc[test_idx]\n",
    "            ###########################################\n",
    "            y_train = np.array(y_train).reshape(-1,1) # 1열짜리로 만드는 것\n",
    "            y_test = np.array(y_test).reshape(-1,1)\n",
    "\n",
    "            ######################################################## cv를 위한 데이터 slicing 작업 끝\n",
    "\n",
    "            # fitting은 cv 안에 넣기\n",
    "            if model_nm == 'LGBM':\n",
    "                model_ = eval(model_nm).fit(X_train, y_train,\n",
    "                                            eval_set=[(X_test, y_test)],\n",
    "                                            eval_metric='auc',\n",
    "                                            early_stopping_rounds=5) # map: mean average precision\n",
    "                # eval_metric = logloss, auc, error\n",
    "                y_pred_test = model_.predict(X_test, num_iteration=LGBM.best_iteration_) # 예측\n",
    "            # If early stopping occurs, the model will have three additional fields: best_score, best_iteration and best_ntree_limit\n",
    "            # [eval_metric](https://stats.stackexchange.com/questions/493981/xgboost-mean-average-precision-eval-metric-for-classification)\n",
    "            # [gbm.fit verbose = 10, #verbose = 10 : 10번 반복할 때마다 logloss값을 보여준다.](https://dacon.io/codeshare/1827)\n",
    "            # [Classifier, Regressor eval_metric 예시](https://hwi-doc.tistory.com/entry/%EC%9D%B4%ED%95%B4%ED%95%98%EA%B3%A0-%EC%82%AC%EC%9A%A9%ED%95%98%EC%9E%90-XGBoost)\n",
    "            # [XGBoost parameters](https://xgboost.readthedocs.io/en/latest/parameter.html#)\n",
    "            # error: For MAP metric, there should be query information\n",
    "            elif model_nm == 'XGBoost': # xgboost_linear가 느려서\n",
    "                cntcnt += 1\n",
    "                print(f'$$$$$---{cntcnt}')\n",
    "                model_ = eval(model_nm).fit(X_train, y_train,\n",
    "                                            eval_set=[(X_test, y_test)],\n",
    "                                            eval_metric='auc',\n",
    "                                            early_stopping_rounds=5) # ,verbose = True\n",
    "                y_pred_test = model_.predict(X_test) # 예측\n",
    "            \n",
    "            else:\n",
    "                cntcnt += 1\n",
    "                print(f'$$$$$---{cntcnt}')\n",
    "                model_ = eval(model_nm).fit(X_train, y_train)\n",
    "                y_pred_test = model_.predict(X_test) # 예측\n",
    "\n",
    "            accuracy = accuracy_score(y_test, y_pred_test)\n",
    "            roc_auc = roc_auc_score(y_test, y_pred_test)\n",
    "            f1 = f1_score(y_test, y_pred_test)\n",
    "            jaccard = jaccard_score(y_test, y_pred_test)\n",
    "#             cv_scores = cross_val_score(eval(model_nm), tmpX, tmpy, cv=7, n_jobs = 64)\n",
    "            \n",
    "            _ = dummy.fit(X_train, y_train)\n",
    "            dummy_score = dummy.score(X_test, y_test)\n",
    "\n",
    "            results.append([i, f'{model_nm}', accuracy, roc_auc, f1, jaccard, dummy_score]) # , cv_scores\n",
    "            i+=1\n",
    "        ################################################################\n",
    "        print(results[-7:])\n",
    "    return results#score_dic#[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c1dbc919",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "  0%|                                                     | 0/5 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---MLP start--->>>\n",
      "14:48:28\n",
      "<<<---1215 start--->>>\n",
      "14:48:28\n",
      "$$$$$---1\n",
      "<<<---1316 start--->>>\n",
      "14:50:06\n",
      "$$$$$---2\n",
      "<<<---1417 start--->>>\n",
      "14:53:03\n",
      "$$$$$---3\n",
      "<<<---1518 start--->>>\n",
      "14:56:02\n",
      "$$$$$---4\n",
      "<<<---1619 start--->>>\n",
      "14:57:35\n",
      "$$$$$---5\n",
      "<<<---1720 start--->>>\n",
      "15:01:35\n",
      "$$$$$---6\n",
      "<<<---1821 start--->>>\n",
      "15:03:24\n",
      "$$$$$---7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 20%|████████▏                                | 1/5 [16:59<1:07:59, 1019.98s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0, 'MLP', 0.5152102200972767, 0.49761857430182693, 0.34155530496472286, 0.2059491679084647, 0.5057084053047293], [1, 'MLP', 0.5201845269032183, 0.5002333308412108, 0.3265675583574715, 0.1951483371727482, 0.5022553709696388], [2, 'MLP', 0.5439007160962367, 0.5006503630676291, 0.18860113840770767, 0.1041190553923169, 0.5052749190871274], [3, 'MLP', 0.5158476878283642, 0.5005430810842111, 0.3908662268135754, 0.24290474373648735, 0.5037980448608468], [4, 'MLP', 0.5307275402892933, 0.49832344494313824, 0.19476744186046513, 0.10789049919484701, 0.5045655216969958], [5, 'MLP', 0.5126817356053541, 0.5055070517253918, 0.40456461599262283, 0.2535763090426432, 0.5020242260232988], [6, 'MLP', 0.5292255009198031, 0.5016852276472746, 0.2730757973427506, 0.15812842099413743, 0.502433127827773]]\n",
      "<<<---DecisionTree start--->>>\n",
      "15:05:28\n",
      "<<<---1215 start--->>>\n",
      "15:05:28\n",
      "$$$$$---1\n",
      "<<<---1316 start--->>>\n",
      "15:10:18\n",
      "$$$$$---2\n",
      "<<<---1417 start--->>>\n",
      "15:14:54\n",
      "$$$$$---3\n",
      "<<<---1518 start--->>>\n",
      "15:19:40\n",
      "$$$$$---4\n",
      "<<<---1619 start--->>>\n",
      "15:24:34\n",
      "$$$$$---5\n",
      "<<<---1720 start--->>>\n",
      "15:28:39\n",
      "$$$$$---6\n",
      "<<<---1821 start--->>>\n",
      "15:32:43\n",
      "$$$$$---7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 40%|████████████████▍                        | 2/5 [49:13<1:17:52, 1557.48s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[7, 'DecisionTree', 0.543082634296239, 0.5181294685684631, 0.2867258922319828, 0.1673555276017785, 0.5038061337125153], [8, 'DecisionTree', 0.5407730857799714, 0.5156496952707778, 0.29042223303418074, 0.1698795098099725, 0.5051841285618358], [9, 'DecisionTree', 0.5514218042300927, 0.518265283120038, 0.3088676748851172, 0.18263956657805297, 0.5043884556606691], [10, 'DecisionTree', 0.5502984759023406, 0.5193972236888679, 0.29056627013045666, 0.16997808400132097, 0.503863107727842], [11, 'DecisionTree', 0.5445174257923262, 0.5022794876340585, 0.02865759380310151, 0.014537095997639275, 0.5030652166110333], [12, 'DecisionTree', 0.5325870740634923, 0.5176533063359036, 0.27669540481582977, 0.16056093948165862, 0.5015606216611053], [13, 'DecisionTree', 0.5453717744742207, 0.5098813605016836, 0.18546335216991902, 0.10220975828275825, 0.503228633222294]]\n",
      "<<<---RandomForest start--->>>\n",
      "15:37:42\n",
      "<<<---1215 start--->>>\n",
      "15:37:42\n",
      "$$$$$---1\n",
      "<<<---1316 start--->>>\n",
      "15:37:56\n",
      "$$$$$---2\n",
      "<<<---1417 start--->>>\n",
      "15:38:12\n",
      "$$$$$---3\n",
      "<<<---1518 start--->>>\n",
      "15:38:26\n",
      "$$$$$---4\n",
      "<<<---1619 start--->>>\n",
      "15:38:40\n",
      "$$$$$---5\n",
      "<<<---1720 start--->>>\n",
      "15:38:54\n",
      "$$$$$---6\n",
      "<<<---1821 start--->>>\n",
      "15:39:08\n",
      "$$$$$---7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 60%|██████████████████████████▍                 | 3/5 [50:58<29:48, 894.19s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[14, 'RandomForest', 0.5391285814716202, 0.5, 0.0, 0.0, 0.5040669803188055], [15, 'RandomForest', 0.5439446419290346, 0.5068852062322965, 0.08797578314075671, 0.04601185610780011, 0.50262715625984], [16, 'RandomForest', 0.5603741614379175, 0.510242172249833, 0.1214419501573154, 0.06464636542239685, 0.5042961157204131], [17, 'RandomForest', 0.5575041884220628, 0.5178050335273706, 0.1870461459385478, 0.1031720391114862, 0.5047804941524748], [18, 'RandomForest', 0.5474426617852912, 0.5084189321166515, 0.10471828735754607, 0.055252096117967046, 0.5044506658052474], [19, 'RandomForest', 0.5212449394349418, 0.5015962547843912, 0.11772244326225281, 0.06254255268616306, 0.5027196325665892], [20, 'RandomForest', 0.5428982498881321, 0.49942829413000756, 0.027451058174821984, 0.013916540975364504, 0.50320998856461]]\n",
      "<<<---GradientBoosting start--->>>\n",
      "15:39:27\n",
      "<<<---1215 start--->>>\n",
      "15:39:27\n",
      "$$$$$---1\n",
      "<<<---1316 start--->>>\n",
      "16:27:17\n",
      "$$$$$---2\n",
      "<<<---1417 start--->>>\n",
      "17:14:40\n",
      "$$$$$---3\n",
      "<<<---1518 start--->>>\n",
      "18:03:46\n",
      "$$$$$---4\n",
      "<<<---1619 start--->>>\n",
      "18:50:17\n",
      "$$$$$---5\n",
      "<<<---1720 start--->>>\n",
      "19:30:51\n",
      "$$$$$---6\n",
      "<<<---1821 start--->>>\n",
      "20:09:17\n",
      "$$$$$---7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 80%|███████████████████████████████▏       | 4/5 [6:11:44<2:15:39, 8139.42s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[21, 'GradientBoosting', 0.5391285814716202, 0.5, 0.0, 0.0, 0.504531414520249], [22, 'GradientBoosting', 0.5403671569427109, 0.5, 0.0, 0.0, 0.503218977742201], [23, 'GradientBoosting', 0.5567498187828672, 0.5, 0.0, 0.0, 0.5029664205807259], [24, 'GradientBoosting', 0.5494266334846045, 0.5019321974697658, 0.03383233115225247, 0.017207246201988234, 0.5024382309406464], [25, 'GradientBoosting', 0.5442410538028067, 0.501804594410495, 0.023381198132580622, 0.011828885827905276, 0.5029934316786906], [26, 'GradientBoosting', 0.5242934076537846, 0.5027626880594303, 0.05038363446576345, 0.025842845472810367, 0.5031886276771804], [27, 'GradientBoosting', 0.5441536568388604, 0.499768369220582, 0.008502700181814496, 0.004269501235372376, 0.504042783274499]]\n",
      "<<<---KNeighbors start--->>>\n",
      "21:00:13\n",
      "<<<---1215 start--->>>\n",
      "21:00:13\n",
      "$$$$$---1\n",
      "<<<---1316 start--->>>\n",
      "22:31:32\n",
      "$$$$$---2\n",
      "<<<---1417 start--->>>\n",
      "23:54:31\n",
      "$$$$$---3\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "start = time.time()\n",
    "_ = my_classifier()\n",
    "results\n",
    "time.time()-start\n",
    "\n",
    "# local variable 변수명 referenced before assignment\n",
    "# 함수 밖에서 설정한 변수를 함수 안에서 쓰려고 할 때 발생하는 에러\n",
    "\n",
    "# logistic regression : 총 22분 걸림 (텀 당 3분 ~ 3분 30초 정도)\n",
    "# 'SVM' : 한 타임에 40분씩 걸림.( 총 280분 (6시간 40분))\n",
    "# 'Gaussian_NB' : 3m\n",
    "# 'KNeighbors' : 1term 3h 25m --> 1h 30m (cv_score 제외 한 후)\n",
    "# -------------------------------------\n",
    "# 'MLP' : 텀 하나에 2~3분, 총 17분.\n",
    "# 'DecisionTree' : 텀 하나에 5분, 총 32분 걸림.\n",
    "# 'RandomForest' : 텀 하나에 14초..? 이거 왜 빠름..? 1분 30초 걸림.\n",
    "# 'GradientBoosting' : 텀 하나에 50분.\n",
    "\n",
    "### cv_scores가 빠져서 그런가 tree기반이 빨라짐. 고로 튜닝 좀더 깊게 해도 될 것 같음.\n",
    "\n",
    "# --------------------------------------\n",
    "# 'XGBoost' : 텀 하나에 8분 (총 1시간 걸림.)\n",
    "# 'LGBM' : # LGBM은 포기;;;\n",
    "# 'dummy' : 1분"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2d16ee7c",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2f738022",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ef62c0bc",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "465b0acd",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "  0%|                                                    | 0/10 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---XGBoost start--->>>\n",
      "13:07:09\n",
      "<<<---1215 start--->>>\n",
      "13:07:09\n",
      "$$$$$---1\n",
      "[0]\tvalidation_0-auc:0.54430\n",
      "[1]\tvalidation_0-auc:0.54672\n",
      "[2]\tvalidation_0-auc:0.54949\n",
      "[3]\tvalidation_0-auc:0.55033\n",
      "[4]\tvalidation_0-auc:0.55229\n",
      "[5]\tvalidation_0-auc:0.55346\n",
      "[6]\tvalidation_0-auc:0.55434\n",
      "[7]\tvalidation_0-auc:0.55481\n",
      "[8]\tvalidation_0-auc:0.55506\n",
      "[9]\tvalidation_0-auc:0.55551\n",
      "<<<---1316 start--->>>\n",
      "13:15:25\n",
      "$$$$$---2\n",
      "[0]\tvalidation_0-auc:0.54311\n",
      "[1]\tvalidation_0-auc:0.54671\n",
      "[2]\tvalidation_0-auc:0.54980\n",
      "[3]\tvalidation_0-auc:0.55129\n",
      "[4]\tvalidation_0-auc:0.55332\n",
      "[5]\tvalidation_0-auc:0.55411\n",
      "[6]\tvalidation_0-auc:0.55497\n",
      "[7]\tvalidation_0-auc:0.55600\n",
      "[8]\tvalidation_0-auc:0.55670\n",
      "[9]\tvalidation_0-auc:0.55734\n",
      "<<<---1417 start--->>>\n",
      "13:23:53\n",
      "$$$$$---3\n",
      "[0]\tvalidation_0-auc:0.54180\n",
      "[1]\tvalidation_0-auc:0.54673\n",
      "[2]\tvalidation_0-auc:0.54873\n",
      "[3]\tvalidation_0-auc:0.55027\n",
      "[4]\tvalidation_0-auc:0.55182\n",
      "[5]\tvalidation_0-auc:0.55321\n",
      "[6]\tvalidation_0-auc:0.55405\n",
      "[7]\tvalidation_0-auc:0.55517\n",
      "[8]\tvalidation_0-auc:0.55597\n",
      "[9]\tvalidation_0-auc:0.55684\n",
      "<<<---1518 start--->>>\n",
      "13:32:14\n",
      "$$$$$---4\n",
      "[0]\tvalidation_0-auc:0.53755\n",
      "[1]\tvalidation_0-auc:0.54192\n",
      "[2]\tvalidation_0-auc:0.54472\n",
      "[3]\tvalidation_0-auc:0.54687\n",
      "[4]\tvalidation_0-auc:0.54778\n",
      "[5]\tvalidation_0-auc:0.54892\n",
      "[6]\tvalidation_0-auc:0.54959\n",
      "[7]\tvalidation_0-auc:0.54963\n",
      "[8]\tvalidation_0-auc:0.54971\n",
      "[9]\tvalidation_0-auc:0.55006\n",
      "<<<---1619 start--->>>\n",
      "13:41:01\n",
      "$$$$$---5\n",
      "[0]\tvalidation_0-auc:0.53891\n",
      "[1]\tvalidation_0-auc:0.54271\n",
      "[2]\tvalidation_0-auc:0.54512\n",
      "[3]\tvalidation_0-auc:0.54627\n",
      "[4]\tvalidation_0-auc:0.54700\n",
      "[5]\tvalidation_0-auc:0.54753\n",
      "[6]\tvalidation_0-auc:0.54830\n",
      "[7]\tvalidation_0-auc:0.54876\n",
      "[8]\tvalidation_0-auc:0.54965\n",
      "[9]\tvalidation_0-auc:0.54944\n",
      "<<<---1720 start--->>>\n",
      "13:49:37\n",
      "$$$$$---6\n",
      "[0]\tvalidation_0-auc:0.52436\n",
      "[1]\tvalidation_0-auc:0.52391\n",
      "[2]\tvalidation_0-auc:0.52370\n",
      "[3]\tvalidation_0-auc:0.52507\n",
      "[4]\tvalidation_0-auc:0.52562\n",
      "[5]\tvalidation_0-auc:0.52693\n",
      "[6]\tvalidation_0-auc:0.52727\n",
      "[7]\tvalidation_0-auc:0.52798\n",
      "[8]\tvalidation_0-auc:0.52835\n",
      "[9]\tvalidation_0-auc:0.52958\n",
      "<<<---1821 start--->>>\n",
      "13:58:03\n",
      "$$$$$---7\n",
      "[0]\tvalidation_0-auc:0.52780\n",
      "[1]\tvalidation_0-auc:0.53103\n",
      "[2]\tvalidation_0-auc:0.53183\n",
      "[3]\tvalidation_0-auc:0.53262\n",
      "[4]\tvalidation_0-auc:0.53301\n",
      "[5]\tvalidation_0-auc:0.53429\n",
      "[6]\tvalidation_0-auc:0.53448\n",
      "[7]\tvalidation_0-auc:0.53521\n",
      "[8]\tvalidation_0-auc:0.53656\n",
      "[9]\tvalidation_0-auc:0.53676\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 10%|███▊                                  | 1/10 [1:00:06<9:00:58, 3606.45s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0, 'XGBoost', 0.5521772738984798, 0.5239100157990049, 0.2508700418258639, 0.14342561606327958, array([0.56581258, 0.56403683, 0.56650351, 0.5674269 , 0.56070488,\n",
      "       0.56339756, 0.56163189]), 0.5057084053047293], [1, 'XGBoost', 0.554286342960549, 0.5265651410217371, 0.27422951710845755, 0.1589026581616947, array([0.5649944 , 0.56230825, 0.55958259, 0.56277569, 0.56259135,\n",
      "       0.56688108, 0.56562358]), 0.5022553709696388], [2, 'XGBoost', 0.5647049046359267, 0.5277673094679406, 0.2918071945256105, 0.17082801987599489, array([0.56005724, 0.56075252, 0.56343238, 0.56495795, 0.56894061,\n",
      "       0.56615275, 0.55919698]), 0.5052749190871274], [3, 'XGBoost', 0.5546316628442233, 0.5247926666951376, 0.30864247406853784, 0.18248210052369776, array([0.56066573, 0.5612311 , 0.56311142, 0.56321306, 0.55555838,\n",
      "       0.57119807, 0.5695782 ]), 0.5037980448608468], [4, 'XGBoost', 0.5541617314525681, 0.5225999440473098, 0.24485230194965074, 0.1395052406254113, array([0.56161982, 0.56300513, 0.55714877, 0.57246591, 0.57370021,\n",
      "       0.56858947, 0.56711881]), 0.5045655216969958], [5, 'XGBoost', 0.5308027363438866, 0.5145795247160921, 0.2405734254141236, 0.13673399554667315, array([0.54730463, 0.55977645, 0.56251417, 0.55252092, 0.57028338,\n",
      "       0.59886554, 0.60045211]), 0.5020242260232988], [6, 'XGBoost', 0.5514033212350221, 0.5162459698921552, 0.20077728812808643, 0.11159112588079634, array([0.55805768, 0.55988045, 0.55016449, 0.59027413, 0.59496519,\n",
      "       0.56899194, 0.55450823]), 0.502433127827773]]\n",
      "<<<---dummy start--->>>\n",
      "14:07:15\n",
      "<<<---1215 start--->>>\n",
      "14:07:15\n",
      "$$$$$---1\n",
      "<<<---1316 start--->>>\n",
      "14:07:29\n",
      "$$$$$---2\n",
      "<<<---1417 start--->>>\n",
      "14:07:40\n",
      "$$$$$---3\n",
      "<<<---1518 start--->>>\n",
      "14:07:50\n",
      "$$$$$---4\n",
      "<<<---1619 start--->>>\n",
      "14:08:01\n",
      "$$$$$---5\n",
      "<<<---1720 start--->>>\n",
      "14:08:11\n",
      "$$$$$---6\n",
      "<<<---1821 start--->>>\n",
      "14:08:23\n",
      "$$$$$---7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 20%|███████▌                              | 2/10 [1:01:26<3:24:15, 1531.97s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[7, 'dummy', 0.5038061337125153, 0.499840406552872, 0.4548568153382681, 0.29437842385969354, array([0.5035967 , 0.50280892, 0.50165952, 0.5059859 , 0.50803931,\n",
      "       0.50574698, 0.50583419]), 0.5038061337125153], [8, 'dummy', 0.5051841285618358, 0.5015875227581409, 0.4591916276909809, 0.2980199458566331, array([0.50164593, 0.50421358, 0.50352229, 0.50283758, 0.50425966,\n",
      "       0.50492794, 0.5037297 ]), 0.5051841285618358], [9, 'dummy', 0.5043884556606691, 0.49978233216229484, 0.4509628417256988, 0.29112461203196194, array([0.50527872, 0.5025246 , 0.50375316, 0.5036384 , 0.50292287,\n",
      "       0.50521797, 0.50335154]), 0.5043884556606691], [10, 'dummy', 0.503863107727842, 0.49956329219324225, 0.45309082041756016, 0.2929007251349196, array([0.50556473, 0.50176598, 0.50451023, 0.50301105, 0.50381146,\n",
      "       0.50396392, 0.50395757]), 0.503863107727842], [11, 'dummy', 0.5030652166110333, 0.4989752389149613, 0.45365470595431173, 0.2933721903517547, array([0.50266228, 0.5048814 , 0.50526219, 0.50257036, 0.50396224,\n",
      "       0.50509481, 0.50282315]), 0.5030652166110333], [12, 'dummy', 0.5015606216611053, 0.4993393207010469, 0.46305458768873403, 0.30128239036960347, array([0.50559472, 0.5031433 , 0.50363239, 0.50518618, 0.50519811,\n",
      "       0.50365926, 0.50213233]), 0.5015606216611053], [13, 'dummy', 0.503228633222294, 0.49994258337227704, 0.4591682426054921, 0.2980002459073983, array([0.50387818, 0.50415186, 0.50396576, 0.50120971, 0.5039083 ,\n",
      "       0.50162025, 0.50195963]), 0.503228633222294]]\n",
      "<<<---Logistic_Regression start--->>>\n",
      "14:08:35\n",
      "<<<---1215 start--->>>\n",
      "14:08:35\n",
      "$$$$$---1\n",
      "<<<---1316 start--->>>\n",
      "14:12:08\n",
      "$$$$$---2\n",
      "[13:08:54] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:17:15] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:25:40] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:34:07] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:42:49] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:51:33] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[14:00:07] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:08:55] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:17:15] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:25:40] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:34:05] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:42:48] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:51:30] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[14:00:10] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:08:56] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:17:17] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:25:38] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:34:06] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:42:50] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:51:31] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[14:00:06] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[13:08:56] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:17:19] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:25:40] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:34:08] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:42:48] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:51:35] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[14:00:08] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:08:58] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:17:18] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:25:40] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:34:05] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:42:47] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:51:39] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[14:00:10] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:08:59] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:17:19] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:25:42] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:34:06] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:42:54] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:51:38] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[14:00:11] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:08:54] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:17:20] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:25:40] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:34:05] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:42:49] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[13:51:33] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n",
      "[14:00:12] WARNING: ../src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/xgboost/sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1417 start--->>>\n",
      "14:15:01\n",
      "$$$$$---3\n",
      "<<<---1518 start--->>>\n",
      "14:18:08\n",
      "$$$$$---4\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
      "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
      "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
      "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
      "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
      "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
      "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
      "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1619 start--->>>\n",
      "14:21:13\n",
      "$$$$$---5\n",
      "<<<---1720 start--->>>\n",
      "14:24:08\n",
      "$$$$$---6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
      "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1821 start--->>>\n",
      "14:27:19\n",
      "$$$$$---7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 30%|███████████▍                          | 3/10 [1:23:31<2:47:43, 1437.63s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[14, 'Logistic_Regression', 0.5176564522952911, 0.4944264227012528, 0.27408776246876226, 0.15880747381501864, array([0.54752557, 0.52997469, 0.52942582, 0.52536419, 0.52675896,\n",
      "       0.52233573, 0.52437961]), 0.5040669803188055], [15, 'Logistic_Regression', 0.5403671569427109, 0.5, 0.0, 0.0, array([0.52321417, 0.54447956, 0.54447297, 0.54447297, 0.54447297,\n",
      "       0.54447656, 0.54447656]), 0.50262715625984], [16, 'Logistic_Regression', 0.5532455180501499, 0.5008853542075404, 0.07278720570338926, 0.03776812084207595, array([0.53795007, 0.53948239, 0.54068394, 0.54071769, 0.54075144,\n",
      "       0.54154122, 0.54328646]), 0.5042961157204131], [17, 'Logistic_Regression', 0.5464142227427251, 0.5001071932728932, 0.052347551858195354, 0.02687725518386946, array([0.54101766, 0.54324736, 0.54215475, 0.54156397, 0.54541354,\n",
      "       0.54310126, 0.54352687]), 0.5047804941524748], [18, 'Logistic_Regression', 0.5416603854850867, 0.49945848872943505, 0.02393946342581977, 0.012114741923505385, array([0.54410326, 0.54498959, 0.54558705, 0.54556735, 0.54560018,\n",
      "       0.54505226, 0.54618809]), 0.5044506658052474], [19, 'Logistic_Regression', 0.5218028818940934, 0.49970645269740815, 0.02122893932539639, 0.010728345359042702, array([0.53946128, 0.54000406, 0.53912131, 0.53937503, 0.54061566,\n",
      "       0.53892173, 0.53995956]), 0.5027196325665892], [20, 'Logistic_Regression', 0.5436751106249689, 0.4997439250246456, 0.018290604004412205, 0.00922971049204882, array([0.53811122, 0.53799627, 0.53831375, 0.53876007, 0.53857396,\n",
      "       0.538043  , 0.53870533]), 0.50320998856461]]\n",
      "<<<---SVM start--->>>\n",
      "14:30:41\n",
      "<<<---1215 start--->>>\n",
      "14:30:41\n",
      "$$$$$---1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/linear_model/_logistic.py:814: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1316 start--->>>\n",
      "15:09:05\n",
      "$$$$$---2\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1417 start--->>>\n",
      "15:46:53\n",
      "$$$$$---3\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1518 start--->>>\n",
      "16:23:48\n",
      "$$$$$---4\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1619 start--->>>\n",
      "17:02:53\n",
      "$$$$$---5\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1720 start--->>>\n",
      "17:40:29\n",
      "$$$$$---6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1821 start--->>>\n",
      "18:22:37\n",
      "$$$$$---7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      " 40%|██████████████▊                      | 4/10 [6:00:17<12:22:35, 7425.97s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[21, 'SVM', 0.4608841427530769, 0.5000062902964115, 0.63094193351698, 0.4608584171581633, array([0.4530685 , 0.45270689, 0.45259712, 0.45259066, 0.45246797,\n",
      "       0.45253254, 0.45251609]), 0.504531414520249], [22, 'SVM', 0.4596328430572891, 0.5, 0.6297924101167254, 0.4596328430572891, array([0.45592205, 0.45567845, 0.45556653, 0.45553361, 0.45556653,\n",
      "       0.45552344, 0.45552344]), 0.503218977742201], [23, 'SVM', 0.4434717970737473, 0.5000535690916323, 0.6139948058269526, 0.4429960352300768, array([0.45588692, 0.45560341, 0.4553469 , 0.45517814, 0.45484063,\n",
      "       0.45494188, 0.4548302 ]), 0.5029664205807259], [24, 'SVM', 0.4512955643390426, 0.5001083295635848, 0.6212842776022687, 0.45062536642564005, array([0.45517723, 0.45504383, 0.45449117, 0.4548342 , 0.45452928,\n",
      "       0.45485961, 0.4544467 ]), 0.5024382309406464], [25, 'SVM', 0.4570044147733391, 0.5001586354834922, 0.6267596946655285, 0.4564093350820143, array([0.45338218, 0.45356601, 0.45338218, 0.45367106, 0.45344127,\n",
      "       0.45347051, 0.45318163]), 0.5029934316786906], [26, 'SVM', 0.47775507673730344, 0.5002212485332145, 0.6457946385480497, 0.47688087562704856, array([0.45982297, 0.45934581, 0.46024049, 0.46008816, 0.45916365,\n",
      "       0.46049971, 0.45934259]), 0.5031886276771804], [27, 'SVM', 0.45577176453040324, 0.500258594839805, 0.6251340435232096, 0.45468726647735763, array([0.46199826, 0.46130309, 0.46159867, 0.46164499, 0.46136583,\n",
      "       0.46172162, 0.46123445]), 0.504042783274499]]\n",
      "<<<---Gaussian_NB start--->>>\n",
      "19:07:27\n",
      "<<<---1215 start--->>>\n",
      "19:07:27\n",
      "$$$$$---1\n",
      "<<<---1316 start--->>>\n",
      "19:08:01\n",
      "$$$$$---2\n",
      "<<<---1417 start--->>>\n",
      "19:08:31\n",
      "$$$$$---3\n",
      "<<<---1518 start--->>>\n",
      "19:09:01\n",
      "$$$$$---4\n",
      "<<<---1619 start--->>>\n",
      "19:09:32\n",
      "$$$$$---5\n",
      "<<<---1720 start--->>>\n",
      "19:10:03\n",
      "$$$$$---6\n",
      "<<<---1821 start--->>>\n",
      "19:10:36\n",
      "$$$$$---7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      " 50%|███████████████████                   | 5/10 [6:04:04<6:42:29, 4830.00s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[28, 'Gaussian_NB', 0.4608777806407283, 0.5000033956225882, 0.6309488294511016, 0.4608657755269537, array([0.45300393, 0.45282312, 0.45261003, 0.45266815, 0.45251317,\n",
      "       0.45251963, 0.45249672]), 0.5045886735313859], [29, 'Gaussian_NB', 0.4596480179670932, 0.5000115749478611, 0.6297912600035869, 0.459631617884174, array([0.45605372, 0.45582329, 0.45565212, 0.45554678, 0.45554019,\n",
      "       0.45552344, 0.45553661]), 0.5031582781029845], [30, 'Gaussian_NB', 0.443924262781002, 0.49983455433994955, 0.6127261677765381, 0.44167644018375757, array([0.45583292, 0.45566416, 0.45522539, 0.45511064, 0.45499588,\n",
      "       0.45504313, 0.45510696]), 0.504111435839901], [31, 'Gaussian_NB', 0.45126628604889474, 0.49998634352508403, 0.6210198524323499, 0.4503472029874772, array([0.455209  , 0.45537416, 0.4544467 , 0.45504383, 0.45466904,\n",
      "       0.45490408, 0.45450387]), 0.5030498218904016], [32, 'Gaussian_NB', 0.456975700800402, 0.500142235900983, 0.6267767893705936, 0.4564274653377837, array([0.45342157, 0.45365793, 0.45352005, 0.45373015, 0.45342157,\n",
      "       0.45345081, 0.45320132]), 0.5043286314202649], [33, 'Gaussian_NB', 0.47804617715077385, 0.5004094175350301, 0.6454424279708807, 0.4764968586783741, array([0.45989455, 0.45943528, 0.46021663, 0.45981975, 0.45897875,\n",
      "       0.46080987, 0.45936048]), 0.502412359907926], [34, 'Gaussian_NB', 0.45682518768955405, 0.5004778353122427, 0.6234392368740791, 0.45289627132649135, array([0.4630328 , 0.46268248, 0.46302733, 0.46238943, 0.46379073,\n",
      "       0.46706407, 0.4625427 ]), 0.5039619897578681]]\n",
      "<<<---KNeighbors start--->>>\n",
      "19:11:14\n",
      "<<<---1215 start--->>>\n",
      "19:11:14\n",
      "$$$$$---1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n",
      "/package/anaconda3.8/lib/python3.8/site-packages/sklearn/svm/_base.py:284: ConvergenceWarning: Solver terminated early (max_iter=1000).  Consider pre-processing your data with StandardScaler or MinMaxScaler.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<<<---1316 start--->>>\n",
      "22:36:06\n",
      "$$$$$---2\n",
      "<<<---1417 start--->>>\n",
      "01:40:33\n",
      "$$$$$---3\n",
      "<<<---1518 start--->>>\n",
      "04:28:21\n",
      "$$$$$---4\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "start = time.time()\n",
    "_ = my_classifier()\n",
    "results\n",
    "time.time()-start\n",
    "\n",
    "# local variable 변수명 referenced before assignment\n",
    "# 함수 밖에서 설정한 변수를 함수 안에서 쓰려고 할 때 발생하는 에러\n",
    "\n",
    "# logistic regression : 총 22분 걸림 (텀 당 3분 ~ 3분 30초 정도)\n",
    "# 'SVM' : 한 타임에 40분씩 걸림.( 총 280분 (6시간 40분))\n",
    "# 'Gaussian_NB' : 3m\n",
    "# 'KNeighbors' : 1term 3h 25m\n",
    "# 'MLP' : \n",
    "# 'DecisionTree' : \n",
    "# 'RandomForest' : \n",
    "# 'GradientBoosting' : \n",
    "# 'XGBoost' : 텀 하나에 8분 (총 1시간 걸림.)\n",
    "# 'LGBM' : # LGBM은 포기;;;\n",
    "# 'dummy' : 1분"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7ff1962f",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0093b118",
   "metadata": {},
   "outputs": [],
   "source": [
    "from datetime import datetime\n",
    "print(datetime.now().strftime('%H:%M:%S'))\n",
    "print(f'<<<---{model_nm} start, {datetime.now().strftime('%H:%M:%S')}--->>>')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "977ceeb8",
   "metadata": {},
   "outputs": [],
   "source": [
    "results = [[0, 'XGBoost', 0.5521772738984798, 0.5239100157990049, 0.2508700418258639, 0.14342561606327958, 0.5057084053047293],\n",
    "           [1, 'XGBoost', 0.554286342960549, 0.5265651410217371, 0.27422951710845755, 0.1589026581616947, 0.5022553709696388],\n",
    "           [2, 'XGBoost', 0.5647049046359267, 0.5277673094679406, 0.2918071945256105, 0.17082801987599489, 0.5052749190871274],\n",
    "           [3, 'XGBoost', 0.5546316628442233, 0.5247926666951376, 0.30864247406853784, 0.18248210052369776, 0.5037980448608468],\n",
    "           [4, 'XGBoost', 0.5541617314525681, 0.5225999440473098, 0.24485230194965074, 0.1395052406254113, 0.5045655216969958],\n",
    "           [5, 'XGBoost', 0.5308027363438866, 0.5145795247160921, 0.2405734254141236, 0.13673399554667315, 0.5020242260232988],\n",
    "           [6, 'XGBoost', 0.5514033212350221, 0.5162459698921552, 0.20077728812808643, 0.11159112588079634, 0.502433127827773],\n",
    "           [7, 'dummy', 0.5038061337125153, 0.499840406552872, 0.4548568153382681, 0.29437842385969354, 0.5038061337125153],\n",
    "           [8, 'dummy', 0.5051841285618358, 0.5015875227581409, 0.4591916276909809, 0.2980199458566331, 0.5051841285618358],\n",
    "           [9, 'dummy', 0.5043884556606691, 0.49978233216229484, 0.4509628417256988, 0.29112461203196194, 0.5043884556606691],\n",
    "           [10, 'dummy', 0.503863107727842, 0.49956329219324225, 0.45309082041756016, 0.2929007251349196, 0.503863107727842],\n",
    "           [11, 'dummy', 0.5030652166110333, 0.4989752389149613, 0.45365470595431173, 0.2933721903517547, 0.5030652166110333],\n",
    "           [12, 'dummy', 0.5015606216611053, 0.4993393207010469, 0.46305458768873403, 0.30128239036960347, 0.5015606216611053],\n",
    "           [13, 'dummy', 0.503228633222294, 0.49994258337227704, 0.4591682426054921, 0.2980002459073983, 0.503228633222294],\n",
    "           [14, 'Logistic_Regression', 0.5176564522952911, 0.4944264227012528, 0.27408776246876226, 0.15880747381501864, 0.5040669803188055],\n",
    "           [15, 'Logistic_Regression', 0.5403671569427109, 0.5, 0.0, 0.0, 0.50262715625984],\n",
    "           [16, 'Logistic_Regression', 0.5532455180501499, 0.5008853542075404, 0.07278720570338926, 0.03776812084207595, 0.5042961157204131],\n",
    "           [17, 'Logistic_Regression', 0.5464142227427251, 0.5001071932728932, 0.052347551858195354, 0.02687725518386946, 0.5047804941524748],\n",
    "           [18, 'Logistic_Regression', 0.5416603854850867, 0.49945848872943505, 0.02393946342581977, 0.012114741923505385, 0.5044506658052474],\n",
    "           [19, 'Logistic_Regression', 0.5218028818940934, 0.49970645269740815, 0.02122893932539639, 0.010728345359042702, 0.5027196325665892],\n",
    "           [20, 'Logistic_Regression', 0.5436751106249689, 0.4997439250246456, 0.018290604004412205, 0.00922971049204882, 0.50320998856461],\n",
    "           [21, 'SVM', 0.4608841427530769, 0.5000062902964115, 0.63094193351698, 0.4608584171581633, 0.504531414520249],\n",
    "           [22, 'SVM', 0.4596328430572891, 0.5, 0.6297924101167254, 0.4596328430572891, 0.503218977742201],\n",
    "           [23, 'SVM', 0.4434717970737473, 0.5000535690916323, 0.6139948058269526, 0.4429960352300768, 0.5029664205807259],\n",
    "           [24, 'SVM', 0.4512955643390426, 0.5001083295635848, 0.6212842776022687, 0.45062536642564005, 0.5024382309406464],\n",
    "           [25, 'SVM', 0.4570044147733391, 0.5001586354834922, 0.6267596946655285, 0.4564093350820143, 0.5029934316786906],\n",
    "           [26, 'SVM', 0.47775507673730344, 0.5002212485332145, 0.6457946385480497, 0.47688087562704856, 0.5031886276771804],\n",
    "           [27, 'SVM', 0.45577176453040324, 0.500258594839805, 0.6251340435232096, 0.45468726647735763, 0.504042783274499],\n",
    "           [28, 'Gaussian_NB', 0.4608777806407283, 0.5000033956225882, 0.6309488294511016, 0.4608657755269537, 0.5045886735313859],\n",
    "           [29, 'Gaussian_NB', 0.4596480179670932, 0.5000115749478611, 0.6297912600035869, 0.459631617884174, 0.5031582781029845],\n",
    "           [30, 'Gaussian_NB', 0.443924262781002, 0.49983455433994955, 0.6127261677765381, 0.44167644018375757, 0.504111435839901],\n",
    "           [31, 'Gaussian_NB', 0.45126628604889474, 0.49998634352508403, 0.6210198524323499, 0.4503472029874772, 0.5030498218904016],\n",
    "           [32, 'Gaussian_NB', 0.456975700800402, 0.500142235900983, 0.6267767893705936, 0.4564274653377837, 0.5043286314202649],\n",
    "           [33, 'Gaussian_NB', 0.47804617715077385, 0.5004094175350301, 0.6454424279708807, 0.4764968586783741, 0.502412359907926],\n",
    "           [34, 'Gaussian_NB', 0.45682518768955405, 0.5004778353122427, 0.6234392368740791, 0.45289627132649135, 0.5039619897578681],\n",
    "           [0, 'MLP', 0.5152102200972767, 0.49761857430182693, 0.34155530496472286, 0.2059491679084647, 0.5057084053047293],\n",
    "           [1, 'MLP', 0.5201845269032183, 0.5002333308412108, 0.3265675583574715, 0.1951483371727482, 0.5022553709696388],\n",
    "           [2, 'MLP', 0.5439007160962367, 0.5006503630676291, 0.18860113840770767, 0.1041190553923169, 0.5052749190871274],\n",
    "           [3, 'MLP', 0.5158476878283642, 0.5005430810842111, 0.3908662268135754, 0.24290474373648735, 0.5037980448608468],\n",
    "           [4, 'MLP', 0.5307275402892933, 0.49832344494313824, 0.19476744186046513, 0.10789049919484701, 0.5045655216969958],\n",
    "           [5, 'MLP', 0.5126817356053541, 0.5055070517253918, 0.40456461599262283, 0.2535763090426432, 0.5020242260232988],\n",
    "           [6, 'MLP', 0.5292255009198031, 0.5016852276472746, 0.2730757973427506, 0.15812842099413743, 0.502433127827773],\n",
    "           [7, 'DecisionTree', 0.543082634296239, 0.5181294685684631, 0.2867258922319828, 0.1673555276017785, 0.5038061337125153],\n",
    "           [8, 'DecisionTree', 0.5407730857799714, 0.5156496952707778, 0.29042223303418074, 0.1698795098099725, 0.5051841285618358],\n",
    "           [9, 'DecisionTree', 0.5514218042300927, 0.518265283120038, 0.3088676748851172, 0.18263956657805297, 0.5043884556606691],\n",
    "           [10, 'DecisionTree', 0.5502984759023406, 0.5193972236888679, 0.29056627013045666, 0.16997808400132097, 0.503863107727842],\n",
    "           [11, 'DecisionTree', 0.5445174257923262, 0.5022794876340585, 0.02865759380310151, 0.014537095997639275, 0.5030652166110333],\n",
    "           [12, 'DecisionTree', 0.5325870740634923, 0.5176533063359036, 0.27669540481582977, 0.16056093948165862, 0.5015606216611053],\n",
    "           [13, 'DecisionTree', 0.5453717744742207, 0.5098813605016836, 0.18546335216991902, 0.10220975828275825, 0.503228633222294],\n",
    "           [14, 'RandomForest', 0.5391285814716202, 0.5, 0.0, 0.0, 0.5040669803188055],\n",
    "           [15, 'RandomForest', 0.5439446419290346, 0.5068852062322965, 0.08797578314075671, 0.04601185610780011, 0.50262715625984],\n",
    "           [16, 'RandomForest', 0.5603741614379175, 0.510242172249833, 0.1214419501573154, 0.06464636542239685, 0.5042961157204131],\n",
    "           [17, 'RandomForest', 0.5575041884220628, 0.5178050335273706, 0.1870461459385478, 0.1031720391114862, 0.5047804941524748],\n",
    "           [18, 'RandomForest', 0.5474426617852912, 0.5084189321166515, 0.10471828735754607, 0.055252096117967046, 0.5044506658052474],\n",
    "           [19, 'RandomForest', 0.5212449394349418, 0.5015962547843912, 0.11772244326225281, 0.06254255268616306, 0.5027196325665892],\n",
    "           [20, 'RandomForest', 0.5428982498881321, 0.49942829413000756, 0.027451058174821984, 0.013916540975364504, 0.50320998856461],\n",
    "           [21, 'GradientBoosting', 0.5391285814716202, 0.5, 0.0, 0.0, 0.504531414520249],\n",
    "           [22, 'GradientBoosting', 0.5403671569427109, 0.5, 0.0, 0.0, 0.503218977742201],\n",
    "           [23, 'GradientBoosting', 0.5567498187828672, 0.5, 0.0, 0.0, 0.5029664205807259],\n",
    "           [24, 'GradientBoosting', 0.5494266334846045, 0.5019321974697658, 0.03383233115225247, 0.017207246201988234, 0.5024382309406464],\n",
    "           [25, 'GradientBoosting', 0.5442410538028067, 0.501804594410495, 0.023381198132580622, 0.011828885827905276, 0.5029934316786906],\n",
    "           [26, 'GradientBoosting', 0.5242934076537846, 0.5027626880594303, 0.05038363446576345, 0.025842845472810367, 0.5031886276771804],\n",
    "           [27, 'GradientBoosting', 0.5441536568388604, 0.499768369220582, 0.008502700181814496, 0.004269501235372376, 0.504042783274499]\n",
    "           ]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "53067bcc",
   "metadata": {},
   "source": [
    "# plot을 위한 준비"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "6b591239",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('SVM', 0.45797365760917164),\n",
       " ('Gaussian_NB', 0.4582233447254926),\n",
       " ('dummy', 0.5035851853081851),\n",
       " ('MLP', 0.5239682753913637),\n",
       " ('Logistic_Regression', 0.5378316754335751),\n",
       " ('GradientBoosting', 0.5426229012824649),\n",
       " ('DecisionTree', 0.5440074677912404),\n",
       " ('RandomForest', 0.5446482034812857),\n",
       " ('XGBoost', 0.5517382819100937)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "[('Logistic_Regression', 0.49918969094759646),\n",
       " ('dummy', 0.49986152809354795),\n",
       " ('SVM', 0.5001152382583057),\n",
       " ('Gaussian_NB', 0.5001236224548198),\n",
       " ('MLP', 0.5006515819443832),\n",
       " ('GradientBoosting', 0.5008954070228963),\n",
       " ('RandomForest', 0.5063394132915072),\n",
       " ('DecisionTree', 0.5144651178742561),\n",
       " ('XGBoost', 0.5223515102341968)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "[('GradientBoosting', 0.01658569484748729),\n",
       " ('Logistic_Regression', 0.06609736096942502),\n",
       " ('RandomForest', 0.09233652400446297),\n",
       " ('DecisionTree', 0.2381997744386554),\n",
       " ('XGBoost', 0.2588217490029044),\n",
       " ('MLP', 0.30285686910561654),\n",
       " ('dummy', 0.4562828059172923),\n",
       " ('Gaussian_NB', 0.62716350912559),\n",
       " ('SVM', 0.6276716862571021)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "[('GradientBoosting', 0.008449782676868035),\n",
       " ('Logistic_Regression', 0.03650366394508014),\n",
       " ('RandomForest', 0.049363064345882536),\n",
       " ('DecisionTree', 0.13816578310759728),\n",
       " ('XGBoost', 0.14906696523964966),\n",
       " ('MLP', 0.181102361920235),\n",
       " ('dummy', 0.29558264764456643),\n",
       " ('Gaussian_NB', 0.4569059474178588),\n",
       " ('SVM', 0.45744144843679857)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "[('SVM', 0.5033399837734561),\n",
       " ('GradientBoosting', 0.5033399837734561),\n",
       " ('DecisionTree', 0.5035851853081851),\n",
       " ('dummy', 0.5035851853081851),\n",
       " ('Gaussian_NB', 0.5036587414929617),\n",
       " ('MLP', 0.5037228022529157),\n",
       " ('XGBoost', 0.5037228022529157),\n",
       " ('Logistic_Regression', 0.5037358619125686),\n",
       " ('RandomForest', 0.5037358619125686)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_list = ['Logistic_Regression', 'SVM', 'Gaussian_NB',\n",
    "              'MLP',\n",
    "              'DecisionTree', 'RandomForest', 'GradientBoosting',\n",
    "              'XGBoost', 'dummy'] #  'LGBM', 'KNeighbors',\n",
    "acc_dic = {i:[] for i in model_list}\n",
    "rocauc_dic = {i:[] for i in model_list}\n",
    "f1_dic = {i:[] for i in model_list}\n",
    "jaccard_dic = {i:[] for i in model_list}\n",
    "dum_dic = {i:[] for i in model_list}\n",
    "for i in results:\n",
    "    acc_dic[i[1]].append(i[2])\n",
    "    rocauc_dic[i[1]].append(i[3])\n",
    "    f1_dic[i[1]].append(i[4])\n",
    "    jaccard_dic[i[1]].append(i[5])\n",
    "    dum_dic[i[1]].append(i[6])\n",
    "\n",
    "\n",
    "# import math\n",
    "# 리스트에 mean 함수가 내장이 안돼있다니...\n",
    "mean_acc_score = {i:[] for i in model_list}\n",
    "mean_rocauc_score = {i:[] for i in model_list}\n",
    "mean_f1_score = {i:[] for i in model_list}\n",
    "mean_jaccard_score = {i:[] for i in model_list}\n",
    "mean_dum_score = {i:[] for i in model_list}\n",
    "for i in model_list:\n",
    "    mean_acc_score[i] = sum(acc_dic[i])/len(acc_dic[i])\n",
    "    mean_rocauc_score[i] = sum(rocauc_dic[i])/len(rocauc_dic[i])\n",
    "    mean_f1_score[i] = sum(f1_dic[i])/len(f1_dic[i])\n",
    "    mean_jaccard_score[i] = sum(jaccard_dic[i])/len(jaccard_dic[i])\n",
    "    mean_dum_score[i] = sum(dum_dic[i])/len(dum_dic[i])\n",
    "\n",
    "\n",
    "# 정렬\n",
    "acc_score_lst = sorted(mean_acc_score.items(), key = lambda t : t[1]) #acc 작은 순으로 줄 세워짐.\n",
    "rocauc_score_lst = sorted(mean_rocauc_score.items(), key = lambda t : t[1]) #rocauc 작은 순으로 줄 세워짐.\n",
    "f1_score_lst = sorted(mean_f1_score.items(), key = lambda t : t[1]) #fi 작은 순으로 줄 세워짐.\n",
    "jaccard_score_lst = sorted(mean_jaccard_score.items(), key = lambda t : t[1])\n",
    "dum_score_lst = sorted(mean_dum_score.items(), key = lambda t : t[1])\n",
    "\n",
    "acc_score_lst\n",
    "rocauc_score_lst\n",
    "f1_score_lst\n",
    "jaccard_score_lst\n",
    "dum_score_lst"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "aee9251a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 다시 dic\n",
    "acc_score_dic = {}\n",
    "rocauc_score_dic = {}\n",
    "f1_score_dic = {}\n",
    "jaccard_score_dic = {}\n",
    "dum_score_dic = {}\n",
    "\n",
    "for i in acc_score_lst:\n",
    "    acc_score_dic[i[0]] = i[1]\n",
    "rocauc_score_dic = {i[0]:i[1] for i in rocauc_score_lst}\n",
    "f1_score_dic = {i[0]:i[1] for i in f1_score_lst}\n",
    "jaccard_score_dic = {i[0]:i[1] for i in jaccard_score_lst}\n",
    "dum_score_dic = {i[0]:i[1] for i in dum_score_lst}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "f949f15a",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>SVM</td>\n",
       "      <td>0.457974</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Gaussian_NB</td>\n",
       "      <td>0.458223</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>dummy</td>\n",
       "      <td>0.503585</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>MLP</td>\n",
       "      <td>0.523968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Logistic_Regression</td>\n",
       "      <td>0.537832</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>GradientBoosting</td>\n",
       "      <td>0.542623</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>DecisionTree</td>\n",
       "      <td>0.544007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>RandomForest</td>\n",
       "      <td>0.544648</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>XGBoost</td>\n",
       "      <td>0.551738</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   key     value\n",
       "0                  SVM  0.457974\n",
       "1          Gaussian_NB  0.458223\n",
       "2                dummy  0.503585\n",
       "3                  MLP  0.523968\n",
       "4  Logistic_Regression  0.537832\n",
       "5     GradientBoosting  0.542623\n",
       "6         DecisionTree  0.544007\n",
       "7         RandomForest  0.544648\n",
       "8              XGBoost  0.551738"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logistic_Regression</td>\n",
       "      <td>0.499190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>dummy</td>\n",
       "      <td>0.499862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>SVM</td>\n",
       "      <td>0.500115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Gaussian_NB</td>\n",
       "      <td>0.500124</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MLP</td>\n",
       "      <td>0.500652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>GradientBoosting</td>\n",
       "      <td>0.500895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>RandomForest</td>\n",
       "      <td>0.506339</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>DecisionTree</td>\n",
       "      <td>0.514465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>XGBoost</td>\n",
       "      <td>0.522352</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   key     value\n",
       "0  Logistic_Regression  0.499190\n",
       "1                dummy  0.499862\n",
       "2                  SVM  0.500115\n",
       "3          Gaussian_NB  0.500124\n",
       "4                  MLP  0.500652\n",
       "5     GradientBoosting  0.500895\n",
       "6         RandomForest  0.506339\n",
       "7         DecisionTree  0.514465\n",
       "8              XGBoost  0.522352"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GradientBoosting</td>\n",
       "      <td>0.016586</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Logistic_Regression</td>\n",
       "      <td>0.066097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>RandomForest</td>\n",
       "      <td>0.092337</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>DecisionTree</td>\n",
       "      <td>0.238200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>XGBoost</td>\n",
       "      <td>0.258822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>MLP</td>\n",
       "      <td>0.302857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>dummy</td>\n",
       "      <td>0.456283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Gaussian_NB</td>\n",
       "      <td>0.627164</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>SVM</td>\n",
       "      <td>0.627672</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   key     value\n",
       "0     GradientBoosting  0.016586\n",
       "1  Logistic_Regression  0.066097\n",
       "2         RandomForest  0.092337\n",
       "3         DecisionTree  0.238200\n",
       "4              XGBoost  0.258822\n",
       "5                  MLP  0.302857\n",
       "6                dummy  0.456283\n",
       "7          Gaussian_NB  0.627164\n",
       "8                  SVM  0.627672"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>GradientBoosting</td>\n",
       "      <td>0.008450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Logistic_Regression</td>\n",
       "      <td>0.036504</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>RandomForest</td>\n",
       "      <td>0.049363</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>DecisionTree</td>\n",
       "      <td>0.138166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>XGBoost</td>\n",
       "      <td>0.149067</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>MLP</td>\n",
       "      <td>0.181102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>dummy</td>\n",
       "      <td>0.295583</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Gaussian_NB</td>\n",
       "      <td>0.456906</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>SVM</td>\n",
       "      <td>0.457441</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   key     value\n",
       "0     GradientBoosting  0.008450\n",
       "1  Logistic_Regression  0.036504\n",
       "2         RandomForest  0.049363\n",
       "3         DecisionTree  0.138166\n",
       "4              XGBoost  0.149067\n",
       "5                  MLP  0.181102\n",
       "6                dummy  0.295583\n",
       "7          Gaussian_NB  0.456906\n",
       "8                  SVM  0.457441"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>SVM</td>\n",
       "      <td>0.503340</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>GradientBoosting</td>\n",
       "      <td>0.503340</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>DecisionTree</td>\n",
       "      <td>0.503585</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>dummy</td>\n",
       "      <td>0.503585</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Gaussian_NB</td>\n",
       "      <td>0.503659</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>MLP</td>\n",
       "      <td>0.503723</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>XGBoost</td>\n",
       "      <td>0.503723</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Logistic_Regression</td>\n",
       "      <td>0.503736</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>RandomForest</td>\n",
       "      <td>0.503736</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   key     value\n",
       "0                  SVM  0.503340\n",
       "1     GradientBoosting  0.503340\n",
       "2         DecisionTree  0.503585\n",
       "3                dummy  0.503585\n",
       "4          Gaussian_NB  0.503659\n",
       "5                  MLP  0.503723\n",
       "6              XGBoost  0.503723\n",
       "7  Logistic_Regression  0.503736\n",
       "8         RandomForest  0.503736"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "acc_df = pd.DataFrame(acc_score_dic.items(), columns=['key', 'value'])\n",
    "acc_df\n",
    "\n",
    "rocauc_df = pd.DataFrame(rocauc_score_dic.items(), columns=['key', 'value'])\n",
    "rocauc_df\n",
    "\n",
    "f1_df = pd.DataFrame(f1_score_dic.items(), columns=['key', 'value'])\n",
    "f1_df\n",
    "\n",
    "jaccard_df = pd.DataFrame(jaccard_score_dic.items(), columns=['key', 'value'])\n",
    "jaccard_df\n",
    "\n",
    "dum_df = pd.DataFrame(dum_score_dic.items(), columns=['key', 'value'])\n",
    "dum_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "16e36424",
   "metadata": {},
   "source": [
    "# plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "808512c4",
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b1abce7b",
   "metadata": {},
   "source": [
    "## acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "d0fbbe83",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 750x450 with 0 Axes>"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 750x450 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,9), dpi = 50)\n",
    "# colors = sns.color_palette('hls',len(model_list))\n",
    "g = sns.barplot(x = 'key', y = 'value', data=acc_df, capsize=.2) # , color = colors\n",
    "_ = g.set_xticks(range(len(acc_df)))\n",
    "_ = g.set_xticklabels(g.get_xticklabels(), rotation=35, fontsize = 20)\n",
    "_ = plt.setp(g.get_yticklabels(), fontsize =18)\n",
    "_ = g.set_xlabel(\"classification models\", fontsize = 28)\n",
    "_ = g.set_ylabel(\"Accuracy\", fontsize = 25)\n",
    "_ = g.set_ylim([0.43, 0.57]) # ax.set(ylim=(0.4, 0.6))\n",
    "\n",
    "dummy_mean = acc_df[acc_df.key=='dummy']['value'].values\n",
    "_ = g.plot([-1,len(acc_df)], [dummy_mean, dummy_mean], color = 'red', linewidth = 2.3)\n",
    "_ = g.annotate(text = f'dummy\\n: {dummy_mean[0]:.4f}', xy = (2, dummy_mean), xytext = (0, 0.53),\n",
    "               arrowprops = dict(facecolor = 'red', shrink = 0), fontsize = 18)\n",
    "\n",
    "val = acc_df['value']\n",
    "cnt = 0\n",
    "for i in range(len(acc_df)):\n",
    "    tmp_fval = f'{val[i]:.2f}'\n",
    "    if tmp_fval.index('.') == 3:\n",
    "        _ = g.text(cnt-0.355, 0.432, tmp_fval, fontsize=15)\n",
    "    elif tmp_fval.index('.') == 2:\n",
    "        _ = g.text(cnt-0.285, 0.432, tmp_fval, fontsize=15)\n",
    "    else:\n",
    "        _ = g.text(cnt-0.185, 0.432, tmp_fval, fontsize=15)\n",
    "    cnt += 1\n",
    "\n",
    "plt.tight_layout()\n",
    "# plt.savefig('[14w] -5 -1 acc_cla_plot.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "92899ce2",
   "metadata": {},
   "source": [
    "[Demonstration of multi-metric evaluation on cross_val_score and GridSearchCV](https://scikit-learn.org/stable/auto_examples/model_selection/plot_multi_metric_evaluation.html?highlight=decisiontreeclassifier) <br>\n",
    "[Classifier comparison](https://scikit-learn.org/stable/auto_examples/classification/plot_classifier_comparison.html?highlight=decisiontreeclassifier) <br>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "458b8705",
   "metadata": {},
   "source": [
    "## ROC AUC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "c554a0db",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 750x450 with 0 Axes>"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 750x450 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,9), dpi = 50)\n",
    "g = sns.barplot(x = 'key', y = 'value', data=rocauc_df, capsize=.2)\n",
    "_ = g.set_xticks(range(len(rocauc_df)))\n",
    "_ = g.set_xticklabels(g.get_xticklabels(), rotation=35, fontsize = 20)\n",
    "_ = plt.setp(g.get_yticklabels(), fontsize =18)\n",
    "_ = g.set_xlabel(\"classification models\", fontsize = 28)\n",
    "_ = g.set_ylabel(\"ROC-AUC\", fontsize = 25)\n",
    "_ = g.set_ylim([0.49, 0.525]) # g.set_ylim([0, 18]) # ax.set(ylim=(0.4, 0.6))\n",
    "\n",
    "dummy_mean = rocauc_df[rocauc_df.key=='dummy']['value'].values\n",
    "_ = g.plot([-1,len(rocauc_df)], [dummy_mean, dummy_mean], color = 'red', linewidth = 2.3)\n",
    "_ = g.annotate(text = f'dummy\\n: {dummy_mean[0]:.4f}', xy = (1, dummy_mean), xytext = (-0.5, 0.502),\n",
    "               arrowprops = dict(facecolor = 'red', shrink = 0), fontsize = 18)\n",
    "\n",
    "\n",
    "val = rocauc_df['value']\n",
    "cnt = 0\n",
    "for i in range(len(rocauc_df)):\n",
    "    tmp_fval = f'{val[i]:.2f}'\n",
    "    if tmp_fval.index('.') == 3:\n",
    "        _ = g.text(cnt-0.355, 16.05, tmp_fval, fontsize=15)\n",
    "    elif tmp_fval.index('.') == 2:\n",
    "        _ = g.text(cnt-0.285, 16.05, tmp_fval, fontsize=15)\n",
    "    else:\n",
    "        _ = g.text(cnt-0.185, 0.4905, tmp_fval, fontsize=15)\n",
    "    cnt += 1\n",
    "\n",
    "plt.tight_layout()\n",
    "# plt.savefig('[14w] -5 -2 rocauc_cla_plot.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4f5904ac",
   "metadata": {},
   "source": [
    "## F1 score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "8afea5cf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 750x450 with 0 Axes>"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 750x450 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,9), dpi = 50)\n",
    "g = sns.barplot(x = 'key', y = 'value', data=f1_df, capsize=.2)\n",
    "_ = g.set_xticks(range(len(f1_df)))\n",
    "_ = g.set_xticklabels(g.get_xticklabels(), rotation=35, fontsize = 20)\n",
    "_ = plt.setp(g.get_yticklabels(), fontsize =18)\n",
    "_ = g.set_xlabel(\"classification models\", fontsize = 28)\n",
    "_ = g.set_ylabel(\"F1 score\", fontsize = 25)\n",
    "_ = g.set_ylim([0, 0.65]) # ax.set(ylim=(0.4, 0.6))\n",
    "\n",
    "dummy_mean = f1_df[f1_df.key=='dummy']['value'].values\n",
    "_ = g.plot([-1,len(f1_df)], [dummy_mean, dummy_mean], color = 'red', linewidth = 2.3)\n",
    "_ = g.annotate(text = f'dummy\\n: {dummy_mean[0]:.4f}', xy = (6, dummy_mean), xytext = (4, 0.5),\n",
    "               arrowprops = dict(facecolor = 'red', shrink = 0), fontsize = 18)\n",
    "\n",
    "\n",
    "val = f1_df['value']\n",
    "cnt = 0\n",
    "for i in range(len(f1_df)):\n",
    "    tmp_fval = f'{val[i]:.2f}'\n",
    "    if tmp_fval.index('.') == 3:\n",
    "        _ = g.text(cnt-0.355, 16.05, tmp_fval, fontsize=15)\n",
    "    elif tmp_fval.index('.') == 2:\n",
    "        _ = g.text(cnt-0.285, 16.05, tmp_fval, fontsize=15)\n",
    "    else:\n",
    "        _ = g.text(cnt-0.185, 0.01, tmp_fval, fontsize=15)\n",
    "    cnt += 1\n",
    "\n",
    "plt.tight_layout()\n",
    "# plt.savefig('[14w] -5 -3 f1_cla_plot.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8527e28e",
   "metadata": {},
   "source": [
    "# Jaccard index\n",
    "희소데이터(데이터 값이 0이 많은 데이터)가 많은 경우에 사용함. <br>\n",
    "\n",
    "[jaccard idex](https://medium.com/h-document/%EC%9E%90%EC%B9%B4%EB%93%9C-%EA%B1%B0%EB%A6%AC-jaccard-distance-e5b246603775)\n",
    "\n",
    "<장점>\n",
    "- 0이 많은 데이터에서는 해당 부분을 고려해줌\n",
    "\n",
    "<단점>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "6fe620cc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 750x450 with 0 Axes>"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 750x450 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,9), dpi = 50)\n",
    "g = sns.barplot(x = 'key', y = 'value', data=jaccard_df, capsize=.2)\n",
    "_ = g.set_xticks(range(len(jaccard_df)))\n",
    "_ = g.set_xticklabels(g.get_xticklabels(), rotation=35, fontsize = 20)\n",
    "_ = plt.setp(g.get_yticklabels(), fontsize =18)\n",
    "_ = g.set_xlabel(\"classification models\", fontsize = 28)\n",
    "_ = g.set_ylabel(\"Jaccard index\", fontsize = 25)\n",
    "_ = g.set_ylim([0, 0.6]) # g.set_ylim([0, 18]) # ax.set(ylim=(0.4, 0.6))\n",
    "\n",
    "dummy_mean = jaccard_df[jaccard_df.key=='dummy']['value'].values\n",
    "_ = g.plot([-1,len(jaccard_df)], [dummy_mean, dummy_mean], color = 'red', linewidth = 2.3)\n",
    "_ = g.annotate(text = f'dummy\\n: {dummy_mean[0]:.4f}', xy = (6, dummy_mean), xytext = (4, 0.35),\n",
    "               arrowprops = dict(facecolor = 'red', shrink = 0), fontsize = 18)\n",
    "\n",
    "\n",
    "val = jaccard_df['value']\n",
    "cnt = 0\n",
    "for i in range(len(jaccard_df)):\n",
    "    tmp_fval = f'{val[i]:.2f}'\n",
    "    if tmp_fval.index('.') == 3:\n",
    "        _ = g.text(cnt-0.355, 16.05, tmp_fval, fontsize=15)\n",
    "    elif tmp_fval.index('.') == 2:\n",
    "        _ = g.text(cnt-0.285, 16.05, tmp_fval, fontsize=15)\n",
    "    else:\n",
    "        _ = g.text(cnt-0.185, 0.01, tmp_fval, fontsize=15)\n",
    "    cnt += 1\n",
    "\n",
    "plt.tight_layout()\n",
    "# plt.savefig('[14w] -5 -4 jaccard_cla_plot.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8fefdaee",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.2"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}