0_format_merge_data.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "from matplotlib import pyplot as plt\n",
    "from matminer.featurizers.base import MultipleFeaturizer, StackedFeaturizer\n",
    "from matminer.featurizers import composition as cf\n",
    "from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor\n",
    "from sklearn.model_selection import GridSearchCV, ShuffleSplit, LeaveOneGroupOut, cross_val_score, learning_curve, KFold\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.preprocessing import Imputer, LabelEncoder\n",
    "from sklearn.metrics import roc_curve, auc, r2_score, make_scorer\n",
    "from sklearn import metrics\n",
    "from pymatgen import Composition\n",
    "import pickle as pkl\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import gzip\n",
    "import os\n",
    "import copy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def featCleanImpute(Y):\n",
    "    ''' Convert inf to NaN in feature array, in place\n",
    "    \n",
    "    Args: \n",
    "        Y, feature array, list of ndarray (#samples x #features)\n",
    "    Returns:\n",
    "        None\n",
    "    '''\n",
    "    # Clean inf values\n",
    "    for i in range(len(Y)):\n",
    "        for j in range(len(Y[i])):\n",
    "            if Y[i][j] == np.inf:\n",
    "                Y[i][j] = np.nan\n",
    "            else:\n",
    "                Y[i][j] = Y[i][j]\n",
    "                \n",
    "    # Impute nan values\n",
    "    imp = Imputer(missing_values='NaN', axis=0, strategy='mean', copy=False)\n",
    "    imp2 = Imputer(missing_values='NaN', axis=1, strategy='mean', copy=False)\n",
    "\n",
    "    imp.fit(Y)\n",
    "    imp.transform(Y)\n",
    "    imp2.fit(Y)\n",
    "    imp.transform(Y)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def genStatsDF(dfOrig):\n",
    "    '''\n",
    "    Generates a new dataframe that merges duplicate entries, while supplying avg/std/no.\n",
    "\n",
    "    input: DataFrame with 2 columns, first is composition, second is property\n",
    "    output: Datafrme with 4 columns\n",
    "\n",
    "    '''\n",
    "    yDcopy = dfOrig.copy()\n",
    "    (col1, col2) = dfOrig.columns\n",
    "\n",
    "    redList = []\n",
    "    cnt=0\n",
    "    while ~yDcopy.empty and cnt < len(yDcopy):\n",
    "        compTemp = yDcopy[col1][0]\n",
    "        \n",
    "        # difference must be below atol+rtol*mean.  \n",
    "        dupes = yDcopy[ [x.almost_equals(compTemp) for x in yDcopy[col1]] ]\n",
    "\n",
    "        tempDict = {col1: compTemp, col2+'_avg': dupes[col2].mean(), \n",
    "                    col2+'_std': dupes[col2].std(), 'numSamp': len(dupes)}\n",
    "\n",
    "        redList.append(tempDict)\n",
    "\n",
    "        #remove processed entries\n",
    "        yDcopy.drop(index=list(dupes.index), inplace=True)\n",
    "        yDcopy.reset_index(drop=True, inplace=True)\n",
    "        cnt+=1\n",
    "\n",
    "    reducedDF = pd.DataFrame(redList)\n",
    "    return reducedDF"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import and Format datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2174"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "elastProp_SLAC = pd.read_excel('datasets/Mechnical properties analysis v33_Pruned.xlsx')\n",
    "rawCopy = elastProp_SLAC.copy()\n",
    "len(rawCopy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>Compositions</th>\n",
       "      <th>Sample</th>\n",
       "      <th>Conditions</th>\n",
       "      <th>Tensile strength (MPa)</th>\n",
       "      <th>Error</th>\n",
       "      <th>Yeild strength, σy (MPa)</th>\n",
       "      <th>Error</th>\n",
       "      <th>Elongation (%)</th>\n",
       "      <th>Compressive fracture strength, σf (MPa)</th>\n",
       "      <th>Error.1</th>\n",
       "      <th>...</th>\n",
       "      <th>Vd (%)</th>\n",
       "      <th>Strain hardening rate</th>\n",
       "      <th>ρ (g/cm3)</th>\n",
       "      <th>Stress Intensity, Kc</th>\n",
       "      <th>G/K</th>\n",
       "      <th>θd</th>\n",
       "      <th>Energy of fracture,Gc</th>\n",
       "      <th>Glasses</th>\n",
       "      <th>Diameter, mm</th>\n",
       "      <th>Reference2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Zr55Al10Ni5Cu30</td>\n",
       "      <td>Zr-based</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1840</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>B. Guennec et al, Intermetallics 92 (2018) 72–78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Zr61Ti2Cu25Al12</td>\n",
       "      <td>Zr-based</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1600</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1688</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Z-Q Song et al, Acta Materialia 99 (2015) 165–175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Zr56Co24Ag4Al16</td>\n",
       "      <td>Zr-based</td>\n",
       "      <td>(electric charge 100 Ag)</td>\n",
       "      <td>2284</td>\n",
       "      <td>63</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>M. Mohammadi Rahvard et al, Journal of Non-Cry...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Zr56Co24Ag4Al16</td>\n",
       "      <td>Zr-based</td>\n",
       "      <td>(electric charge 90 Ag)</td>\n",
       "      <td>2197</td>\n",
       "      <td>48</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>M. Mohammadi Rahvard et al, Journal of Non-Cry...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Al20Co25Zr55</td>\n",
       "      <td>Al-based</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2050</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.1</td>\n",
       "      <td>2200</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>M. Mohammadi Rahvard et al, Journal of Non-Cry...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Compositions    Sample                 Conditions  \\\n",
       "0  Zr55Al10Ni5Cu30  Zr-based                        NaN   \n",
       "1  Zr61Ti2Cu25Al12  Zr-based                        NaN   \n",
       "2  Zr56Co24Ag4Al16  Zr-based   (electric charge 100 Ag)   \n",
       "3  Zr56Co24Ag4Al16  Zr-based    (electric charge 90 Ag)   \n",
       "4     Al20Co25Zr55  Al-based                        NaN   \n",
       "\n",
       "  Tensile strength (MPa) Error  Yeild strength, σy (MPa)  Error  \\\n",
       "0                   1840    NaN                      NaN    NaN   \n",
       "1                   1600    NaN                      NaN    NaN   \n",
       "2                   2284     63                      NaN    NaN   \n",
       "3                   2197     48                      NaN    NaN   \n",
       "4                   2050    NaN                      NaN    NaN   \n",
       "\n",
       "  Elongation (%) Compressive fracture strength, σf (MPa)  Error.1  \\\n",
       "0              2                                     NaN      NaN   \n",
       "1            NaN                                    1688      NaN   \n",
       "2            NaN                                     NaN      NaN   \n",
       "3            NaN                                     NaN      NaN   \n",
       "4            2.1                                    2200      NaN   \n",
       "\n",
       "                         ...                         Vd (%)  \\\n",
       "0                        ...                            NaN   \n",
       "1                        ...                            NaN   \n",
       "2                        ...                            NaN   \n",
       "3                        ...                            NaN   \n",
       "4                        ...                            NaN   \n",
       "\n",
       "   Strain hardening rate ρ (g/cm3) Stress Intensity, Kc G/K   θd  \\\n",
       "0                    NaN       NaN                  NaN NaN  NaN   \n",
       "1                    NaN       NaN                  NaN NaN  NaN   \n",
       "2                    NaN       NaN                  NaN NaN  NaN   \n",
       "3                    NaN       NaN                  NaN NaN  NaN   \n",
       "4                    NaN       NaN                  NaN NaN  NaN   \n",
       "\n",
       "   Energy of fracture,Gc Glasses  Diameter, mm  \\\n",
       "0                    NaN     BMG           NaN   \n",
       "1                    NaN     BMG           NaN   \n",
       "2                    NaN     BMG           NaN   \n",
       "3                    NaN     BMG           NaN   \n",
       "4                    NaN     BMG           NaN   \n",
       "\n",
       "                                          Reference2  \n",
       "0   B. Guennec et al, Intermetallics 92 (2018) 72–78  \n",
       "1  Z-Q Song et al, Acta Materialia 99 (2015) 165–175  \n",
       "2  M. Mohammadi Rahvard et al, Journal of Non-Cry...  \n",
       "3  M. Mohammadi Rahvard et al, Journal of Non-Cry...  \n",
       "4  M. Mohammadi Rahvard et al, Journal of Non-Cry...  \n",
       "\n",
       "[5 rows x 40 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "elastProp_SLAC.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "elastProp_SLAC.rename(index=str, columns={'θd':'theta_d', 'ρ (g/cm3)':'density (g/cm3)', \n",
    "                                          'Yeild strength, σy (MPa)':'Yeild strength (MPa)', \n",
    "                                         'Compressive fracture strength, σf (MPa)': 'Compressive fracture strength (MPa)'},inplace=True)\n",
    "#elastProp_SLAC.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "dfSLAC = elastProp_SLAC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def makeComp(x):\n",
    "    '''Apply Composition() constructor to string x\n",
    "       Return: Composition object, or 0 if not parse-able'''\n",
    "    try:\n",
    "        return Composition(x)\n",
    "    except:\n",
    "        print(x)\n",
    "        return 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mg60Ca30−xTixY10\n",
      "Mg60Li5Cu25−xZnx (Y, Gd)10\n",
      "Mg65−xLixCu25Znx (Y, Gd)10\n",
      "(CaCu)100xMgx\n",
      "La57.6Al17.5(Cu,Ni)24.9\n",
      "La64Al14 (Cu = Ni)22\n",
      "Fused quartz\n",
      "Amorphous carbon\n",
      "Microcrystal glass\n",
      "Breakaway glass\n",
      "Water-white glass\n",
      "Window glass\n",
      "Float glass\n",
      "Ti-glass\n",
      "Borosilicate glass\n",
      "B2O3 glass\n",
      "Glassy sulfur\n",
      "Amorphous Se\n",
      "Fused quartz\n",
      "window glass (SiO2)\n",
      "∥Fe70Mo5Ni5∣P12.5∥C5B2.5∣\n",
      "Toughened glass (SiO2)\n",
      "Amorphous carbon\n",
      "Vit1\n",
      "Float glass\n",
      "Ti-glass\n",
      "Window glass\n",
      "Water-white glass\n",
      "Fused quartz\n",
      "Microcrystal glass\n",
      "Borosilicate glass\n",
      "Carbon glass\n",
      "Mo52Cr14Fel4P12B8\n",
      "Fused silica\n",
      "Window glass\n",
      "Toughened glass\n",
      "Sn 99.9%\n",
      "(Zr54Al17C029)100-0Ta0\n",
      "316L SS\n",
      "Ti-6Al-4V\n",
      "Cortical bone\n",
      "Vitreloy 105\n",
      "FC20+0.4 wt.% B\n",
      "La74Al14(Cu,Ni)12\n",
      "300 M Steel\n",
      "D6AC Steel \n",
      "Ti-6Al-4V \n",
      "2090-T81 Al–Li \n"
     ]
    }
   ],
   "source": [
    "# prints out compositions that cannot be converted into Composition objects\n",
    "dfSLAC['comp'] = dfSLAC['Compositions'].apply(makeComp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remove entries where composition was unable to be read"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2126 entries remaining\n"
     ]
    }
   ],
   "source": [
    "dfSLAC = dfSLAC[dfSLAC['comp']!=0]\n",
    "dfSLAC.reset_index(drop=True, inplace=True)\n",
    "print('{} entries remaining'.format(len(dfSLAC)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "moved 'comp' to first column\n"
     ]
    }
   ],
   "source": [
    "cols = dfSLAC.columns.tolist()\n",
    "dfSLAC = dfSLAC[cols[-1:] + cols[:-1]]\n",
    "print(\"moved 'comp' to first column\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2126"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(dfSLAC['comp']!=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Hikaru\\Anaconda2\\envs\\py36\\lib\\site-packages\\pymatgen\\core\\periodic_table.py:423: UserWarning:\n",
      "\n",
      "No electronegativity for Ar. Setting to NaN. This has no physical meaning, and is mainly done to avoid errors caused by the code expecting a float.\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "2125 entries remaining\n"
     ]
    }
   ],
   "source": [
    "def redComp(x):\n",
    "    try: \n",
    "        return Composition(x.reduced_formula)\n",
    "    except: \n",
    "        print(x)\n",
    "        return 0\n",
    "    \n",
    "dfSLAC['comp'] = dfSLAC['comp'].apply(redComp)\n",
    "dfSLAC = dfSLAC[dfSLAC['comp']!=0]\n",
    "dfSLAC.reset_index(drop=True, inplace=True)\n",
    "print('{} entries remaining'.format(len(dfSLAC)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "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>comp</th>\n",
       "      <th>Compositions</th>\n",
       "      <th>Sample</th>\n",
       "      <th>Conditions</th>\n",
       "      <th>Tensile strength (MPa)</th>\n",
       "      <th>Error</th>\n",
       "      <th>Yeild strength (MPa)</th>\n",
       "      <th>Error</th>\n",
       "      <th>Elongation (%)</th>\n",
       "      <th>Compressive fracture strength (MPa)</th>\n",
       "      <th>...</th>\n",
       "      <th>Vd (%)</th>\n",
       "      <th>Strain hardening rate</th>\n",
       "      <th>density (g/cm3)</th>\n",
       "      <th>Stress Intensity, Kc</th>\n",
       "      <th>G/K</th>\n",
       "      <th>theta_d</th>\n",
       "      <th>Energy of fracture,Gc</th>\n",
       "      <th>Glasses</th>\n",
       "      <th>Diameter, mm</th>\n",
       "      <th>Reference2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>(Zr, Al, Cu, Ni)</td>\n",
       "      <td>Zr55Al10Ni5Cu30</td>\n",
       "      <td>Zr-based</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1840</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>B. Guennec et al, Intermetallics 92 (2018) 72–78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(Zr, Ti, Al, Cu)</td>\n",
       "      <td>Zr61Ti2Cu25Al12</td>\n",
       "      <td>Zr-based</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1600</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1688</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Z-Q Song et al, Acta Materialia 99 (2015) 165–175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>(Zr, Al, Co, Ag)</td>\n",
       "      <td>Zr56Co24Ag4Al16</td>\n",
       "      <td>Zr-based</td>\n",
       "      <td>(electric charge 100 Ag)</td>\n",
       "      <td>2284</td>\n",
       "      <td>63</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>M. Mohammadi Rahvard et al, Journal of Non-Cry...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>(Zr, Al, Co, Ag)</td>\n",
       "      <td>Zr56Co24Ag4Al16</td>\n",
       "      <td>Zr-based</td>\n",
       "      <td>(electric charge 90 Ag)</td>\n",
       "      <td>2197</td>\n",
       "      <td>48</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>M. Mohammadi Rahvard et al, Journal of Non-Cry...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>(Zr, Al, Co)</td>\n",
       "      <td>Al20Co25Zr55</td>\n",
       "      <td>Al-based</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2050</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.1</td>\n",
       "      <td>2200</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BMG</td>\n",
       "      <td>NaN</td>\n",
       "      <td>M. Mohammadi Rahvard et al, Journal of Non-Cry...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 41 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "               comp     Compositions    Sample                 Conditions  \\\n",
       "0  (Zr, Al, Cu, Ni)  Zr55Al10Ni5Cu30  Zr-based                        NaN   \n",
       "1  (Zr, Ti, Al, Cu)  Zr61Ti2Cu25Al12  Zr-based                        NaN   \n",
       "2  (Zr, Al, Co, Ag)  Zr56Co24Ag4Al16  Zr-based   (electric charge 100 Ag)   \n",
       "3  (Zr, Al, Co, Ag)  Zr56Co24Ag4Al16  Zr-based    (electric charge 90 Ag)   \n",
       "4      (Zr, Al, Co)     Al20Co25Zr55  Al-based                        NaN   \n",
       "\n",
       "  Tensile strength (MPa) Error  Yeild strength (MPa)  Error Elongation (%)  \\\n",
       "0                   1840    NaN                  NaN    NaN              2   \n",
       "1                   1600    NaN                  NaN    NaN            NaN   \n",
       "2                   2284     63                  NaN    NaN            NaN   \n",
       "3                   2197     48                  NaN    NaN            NaN   \n",
       "4                   2050    NaN                  NaN    NaN            2.1   \n",
       "\n",
       "  Compressive fracture strength (MPa)  \\\n",
       "0                                 NaN   \n",
       "1                                1688   \n",
       "2                                 NaN   \n",
       "3                                 NaN   \n",
       "4                                2200   \n",
       "\n",
       "                         ...                          Vd (%)  \\\n",
       "0                        ...                             NaN   \n",
       "1                        ...                             NaN   \n",
       "2                        ...                             NaN   \n",
       "3                        ...                             NaN   \n",
       "4                        ...                             NaN   \n",
       "\n",
       "  Strain hardening rate  density (g/cm3) Stress Intensity, Kc G/K theta_d  \\\n",
       "0                   NaN              NaN                  NaN NaN     NaN   \n",
       "1                   NaN              NaN                  NaN NaN     NaN   \n",
       "2                   NaN              NaN                  NaN NaN     NaN   \n",
       "3                   NaN              NaN                  NaN NaN     NaN   \n",
       "4                   NaN              NaN                  NaN NaN     NaN   \n",
       "\n",
       "  Energy of fracture,Gc  Glasses Diameter, mm  \\\n",
       "0                   NaN      BMG          NaN   \n",
       "1                   NaN      BMG          NaN   \n",
       "2                   NaN      BMG          NaN   \n",
       "3                   NaN      BMG          NaN   \n",
       "4                   NaN      BMG          NaN   \n",
       "\n",
       "                                          Reference2  \n",
       "0   B. Guennec et al, Intermetallics 92 (2018) 72–78  \n",
       "1  Z-Q Song et al, Acta Materialia 99 (2015) 165–175  \n",
       "2  M. Mohammadi Rahvard et al, Journal of Non-Cry...  \n",
       "3  M. Mohammadi Rahvard et al, Journal of Non-Cry...  \n",
       "4  M. Mohammadi Rahvard et al, Journal of Non-Cry...  \n",
       "\n",
       "[5 rows x 41 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfSLAC.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import NREL dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# cell about Composition.almost_equal() method experimentation was here\n",
    "# Note that an unreduced chemical formula is not equal in composition"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build individual datasets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generate, clean feature set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "base_featurizer = MultipleFeaturizer([cf.Stoichiometry(), cf.ElementProperty.from_preset(\"magpie\"),\n",
    "                                 cf.ValenceOrbital(props=['avg']), cf.IonProperty(fast=True),\n",
    "                                cf.YangSolidSolution(), cf.AtomicPackingEfficiency()])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Young's Modulus dataset \n",
    "Isolate Youngs Modulus, Composition.\n",
    "Build Features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separate Data into new frame\n",
    "youngsSLAC = pd.DataFrame()\n",
    "youngsSLAC['comp'] = dfSLAC['comp']\n",
    "youngsSLAC['E'] = dfSLAC['Young’s  Modulus, E (GPa)']\n",
    "\n",
    "# Drop missing entries \n",
    "youngsSLAC = youngsSLAC[[(type(x) in (int, float)) for x in youngsSLAC['E']]]\n",
    "youngsSLAC.reset_index(drop=True, inplace=True)\n",
    "youngsSLAC = youngsSLAC[~youngsSLAC['E'].isnull()]\n",
    "youngsSLAC['E'].astype(float)\n",
    "youngsSLAC.reset_index(drop=True, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Hikaru\\Anaconda2\\envs\\py36\\lib\\site-packages\\pymatgen\\core\\periodic_table.py:423: UserWarning:\n",
      "\n",
      "No electronegativity for Ar. Setting to NaN. This has no physical meaning, and is mainly done to avoid errors caused by the code expecting a float.\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1203 = 1203 + 0?\n"
     ]
    }
   ],
   "source": [
    "# Join and sort data (SLAC dataset is complete)\n",
    "youngsData = youngsSLAC #.append(youngsNREL, ignore_index=True)\n",
    "youngsData = youngsData.sort_values('comp').reset_index(drop=True)\n",
    "print('{} = {} + {}?'.format(len(youngsData), len(youngsSLAC), len([])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "comp    (Cu, Ar)\n",
       "E          100.5\n",
       "Name: 69, dtype: object"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "youngsData.loc[69]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                  comp     E\n",
      "1201     (M0+, Al, Cu)  30.9\n",
      "1202  (Glass0+, Si, O)  83.6\n"
     ]
    }
   ],
   "source": [
    "# Drop exceptions actually accepted into composition object.... (T0+, M0+, glass0+)\n",
    "print(youngsData.loc[[1201, 1202]])\n",
    "youngsData.drop([1201, 1202], inplace=True)\n",
    "youngsData.reset_index(inplace=True, drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fad5409b522b4383aa932cb982eece3d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='MultipleFeaturizer', max=1201), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Computed 152 features\n",
      "Wall time: 18min 14s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "X_E = base_featurizer.featurize_many(youngsData['comp'], ignore_errors=True)\n",
    "X_E = np.array(X_E)\n",
    "X_E.astype(float)\n",
    "print('Computed {} features'.format(X_E.shape[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create smaller dataset for quick tests\n",
    "youngsDataSmall = youngsData.sample(500).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "XEsmall = base_featurizer.featurize_many(youngsDataSmall['comp'], ignore_errors=True)\n",
    "XEsmall = np.array(XEsmall)\n",
    "XEsmall.astype(float)\n",
    "print('Computed {} features'.format(XEsmall.shape[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Comp: Zr14 Al4 Co7 .... at row: 481\n",
      "Comp: Zr14 Al4 Co7 .... at row: 482\n"
     ]
    }
   ],
   "source": [
    "# some duplicates exist\n",
    "comp = Composition('Zr14Al4Co7')\n",
    "comp2 = Composition('Zr20Al4Co7') \n",
    "for index, row in youngsData[ [x.almost_equals(comp) for x in youngsData['comp']] ].iterrows():\n",
    "    print('Comp: {} .... at row: {}'.format(row['comp'], index))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(np.where(np.isnan(X_E)))\n",
    "featCleanImpute(X_E)\n",
    "print(np.where(np.isnan(X_E)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "with gzip.open('./datasets/youngs_features.pkl.gz', 'wb') as fp:\n",
    "    pkl.dump(X_E, fp)\n",
    "with gzip.open('./datasets/youngs_data.pkl.gz', 'wb') as fd:\n",
    "    pkl.dump(youngsData, fd)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with gzip.open('./datasets/youngsSmall_features.pkl.gz', 'wb') as fp:\n",
    "    pkl.dump(XEsmall, fp)\n",
    "with gzip.open('./datasets/youngsSmall_data.pkl.gz', 'wb') as fd:\n",
    "    pkl.dump(youngsDataSmall, fd)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Generate new dataframe, reducing duplicates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 5.71 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "youngsReduced = genStatsDF(youngsData)\n",
    "youngsReduced.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x15bbf5f8>"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x15bb9438>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "youngsReduced['numSamp'].plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x15bb91d0>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x15cd67f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "youngsReduced['E_std'].plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Density Model Dataset\n",
    "Use as benchmark for validity of other models?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "densitySLAC = pd.DataFrame()\n",
    "densitySLAC['comp'] = dfSLAC['comp']\n",
    "densitySLAC['density'] = dfSLAC['density (g/cm3)']\n",
    "\n",
    "densitySLAC.dropna(inplace=True)\n",
    "densitySLAC.reset_index(drop=True, inplace=True)\n",
    "densitySLAC = densitySLAC[[(type(x) in (int, float)) for x in densitySLAC['density']]]\n",
    "densitySLAC['density'].astype(float)\n",
    "densitySLAC.reset_index(drop=True, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dropping N/A, non number density values leaves 746 values\n"
     ]
    },
    {
     "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>comp</th>\n",
       "      <th>density</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>(Ce, Al)</td>\n",
       "      <td>3.462</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(Al, Fe, B)</td>\n",
       "      <td>3.539</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>(Zr, Al, Ni)</td>\n",
       "      <td>3.384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>(Zr, Al, Ni)</td>\n",
       "      <td>3.517</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>(Hf, Al, Ni)</td>\n",
       "      <td>2.835</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           comp density\n",
       "0      (Ce, Al)   3.462\n",
       "1   (Al, Fe, B)   3.539\n",
       "2  (Zr, Al, Ni)   3.384\n",
       "3  (Zr, Al, Ni)   3.517\n",
       "4  (Hf, Al, Ni)   2.835"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print('dropping N/A, non number density values leaves {} values'.format(len(densitySLAC)))\n",
    "densitySLAC.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "746 = 746 + 0?\n"
     ]
    }
   ],
   "source": [
    "densityData = densitySLAC #.append(densityNREL, ignore_index=True)\n",
    "print('{} = {} + {}?'.format(len(densityData), len(densitySLAC), len([])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                      comp density\n",
      "94   (T0+, Zr, Cu, Sn, Pd)    6.62\n",
      "273          (M0+, Al, Cu)   6.564\n",
      "326       (Glass0+, Si, O)     2.2\n"
     ]
    }
   ],
   "source": [
    "# Drop exceptions actually accepted into composition object.... (T0+, M0+, glass0+)\n",
    "print(densityData.loc[[94, 273, 326]])\n",
    "densityData.drop([94, 273, 326], inplace=True)\n",
    "densityData.reset_index(inplace=True, drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "X_dens = base_featurizer.featurize_many(densityData['comp'], ignore_errors=True)\n",
    "X_dens = np.array(X_dens)\n",
    "X_dens.astype(float)\n",
    "print('Computed {} features'.format(X_dens.shape[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(np.where(np.isnan(X_dens)))\n",
    "featCleanImpute(X_dens)\n",
    "print(np.where(np.isnan(X_dens)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_dens_raw = base_featurizer.featurize_many(densityData['comp'], ignore_errors=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "iinf = set(np.where(np.isinf(X_dens_raw))[0])\n",
    "print(len(iinf))\n",
    "iinf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ierror = inan.union(iinf)\n",
    "X_dens_trim = np.delete(X_dens_raw, list(ierror), axis=0)\n",
    "print(len(X_dens_trim))\n",
    "densityDataTrim = densityData.drop(axis=0,index=list(ierror))\n",
    "densityDataTrim.reset_index(drop=True, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.where(np.isinf(X_dens_trim))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with gzip.open('./datasets/density_features.pkl.gz', 'wb') as fp:\n",
    "    pkl.dump(X_dens_trim, fp)\n",
    "with gzip.open('./datasets/density_data.pkl.gz', 'wb') as fd:\n",
    "    pkl.dump(densityDataTrim, fd)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Generate Reduced Dataset with some stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%time\n",
    "densReduced = genStatsDF(densityData)\n",
    "densReduced.head()\n",
    "\n",
    "print('combining duplicates removes {} entries'.format(len(densityData)-len(densReduced)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "densReduced['density_std'].plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "densReduced['numSamp'].plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "cnt=0\n",
    "\n",
    "for x in densityData['comp']:\n",
    "    print(x)\n",
    "    cnt+=1\n",
    "    if cnt==30:\n",
    "        break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}