Training_ensemble_models.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import sklearn as sk\n",
    "import ase\n",
    "import ase.io as aio\n",
    "import qml\n",
    "import time\n",
    "import warnings\n",
    "import scipy as sp\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.model_selection import train_test_split\n",
    "from qml.kernels import gaussian_kernel\n",
    "from qml.math import cho_solve\n",
    "import copy\n",
    "%config Completer.use_jedi = False\n",
    "%matplotlib notebook"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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>mol</th>\n",
       "      <th>Eafw</th>\n",
       "      <th>Earev</th>\n",
       "      <th>dErxn</th>\n",
       "      <th>ETS</th>\n",
       "      <th>ERt</th>\n",
       "      <th>EPr</th>\n",
       "      <th>enan</th>\n",
       "      <th>HRt</th>\n",
       "      <th>HPr</th>\n",
       "      <th>HTS</th>\n",
       "      <th>Hafw</th>\n",
       "      <th>dH</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1abp1</td>\n",
       "      <td>8.674063</td>\n",
       "      <td>17.793659</td>\n",
       "      <td>-9.119595</td>\n",
       "      <td>-1.453883e+06</td>\n",
       "      <td>-1.453892e+06</td>\n",
       "      <td>-1.453901e+06</td>\n",
       "      <td>R</td>\n",
       "      <td>-1.453681e+06</td>\n",
       "      <td>-1.453689e+06</td>\n",
       "      <td>-1.453672e+06</td>\n",
       "      <td>8.495217</td>\n",
       "      <td>-7.957442</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1abp2</td>\n",
       "      <td>7.572784</td>\n",
       "      <td>19.457186</td>\n",
       "      <td>-11.884402</td>\n",
       "      <td>-1.453885e+06</td>\n",
       "      <td>-1.453892e+06</td>\n",
       "      <td>-1.453904e+06</td>\n",
       "      <td>R</td>\n",
       "      <td>-1.453681e+06</td>\n",
       "      <td>-1.453691e+06</td>\n",
       "      <td>-1.453673e+06</td>\n",
       "      <td>7.375741</td>\n",
       "      <td>-10.650710</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1abp3</td>\n",
       "      <td>12.316128</td>\n",
       "      <td>22.034368</td>\n",
       "      <td>-9.718239</td>\n",
       "      <td>-1.453878e+06</td>\n",
       "      <td>-1.453891e+06</td>\n",
       "      <td>-1.453900e+06</td>\n",
       "      <td>R</td>\n",
       "      <td>-1.453679e+06</td>\n",
       "      <td>-1.453688e+06</td>\n",
       "      <td>-1.453667e+06</td>\n",
       "      <td>12.109041</td>\n",
       "      <td>-8.383520</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1abp4</td>\n",
       "      <td>9.298435</td>\n",
       "      <td>22.482409</td>\n",
       "      <td>-13.183974</td>\n",
       "      <td>-1.453880e+06</td>\n",
       "      <td>-1.453889e+06</td>\n",
       "      <td>-1.453902e+06</td>\n",
       "      <td>R</td>\n",
       "      <td>-1.453678e+06</td>\n",
       "      <td>-1.453690e+06</td>\n",
       "      <td>-1.453669e+06</td>\n",
       "      <td>9.181712</td>\n",
       "      <td>-11.817877</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1abp5</td>\n",
       "      <td>10.178204</td>\n",
       "      <td>22.937354</td>\n",
       "      <td>-12.759150</td>\n",
       "      <td>-1.453881e+06</td>\n",
       "      <td>-1.453892e+06</td>\n",
       "      <td>-1.453904e+06</td>\n",
       "      <td>R</td>\n",
       "      <td>-1.453680e+06</td>\n",
       "      <td>-1.453692e+06</td>\n",
       "      <td>-1.453670e+06</td>\n",
       "      <td>9.975511</td>\n",
       "      <td>-11.326537</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     mol       Eafw      Earev      dErxn           ETS           ERt  \\\n",
       "0  1abp1   8.674063  17.793659  -9.119595 -1.453883e+06 -1.453892e+06   \n",
       "1  1abp2   7.572784  19.457186 -11.884402 -1.453885e+06 -1.453892e+06   \n",
       "2  1abp3  12.316128  22.034368  -9.718239 -1.453878e+06 -1.453891e+06   \n",
       "3  1abp4   9.298435  22.482409 -13.183974 -1.453880e+06 -1.453889e+06   \n",
       "4  1abp5  10.178204  22.937354 -12.759150 -1.453881e+06 -1.453892e+06   \n",
       "\n",
       "            EPr enan           HRt           HPr           HTS       Hafw  \\\n",
       "0 -1.453901e+06    R -1.453681e+06 -1.453689e+06 -1.453672e+06   8.495217   \n",
       "1 -1.453904e+06    R -1.453681e+06 -1.453691e+06 -1.453673e+06   7.375741   \n",
       "2 -1.453900e+06    R -1.453679e+06 -1.453688e+06 -1.453667e+06  12.109041   \n",
       "3 -1.453902e+06    R -1.453678e+06 -1.453690e+06 -1.453669e+06   9.181712   \n",
       "4 -1.453904e+06    R -1.453680e+06 -1.453692e+06 -1.453670e+06   9.975511   \n",
       "\n",
       "          dH  \n",
       "0  -7.957442  \n",
       "1 -10.650710  \n",
       "2  -8.383520  \n",
       "3 -11.817877  \n",
       "4 -11.326537  "
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.read_csv('./data.csv', index_col=0)\n",
    "\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "data['ase_reactants'] = [aio.read(\n",
    "    'data_react_xyz/' + data.mol.values[i] + data.enan.values[i] + '.xyz') for i in range(len(data))]\n",
    "\n",
    "data['ase_products'] = [aio.read(\n",
    "    'data_prod_xyz/' + data.mol.values[i] + data.enan.values[i] + '.xyz') for i in range(len(data))]\n",
    "\n",
    "mbtypes = np.load('./mbtypes.npy', allow_pickle=True)\n",
    "\n",
    "data['SLATM_react'] = [qml.representations.generate_slatm(coordinates=mol.positions, \n",
    "                                                 nuclear_charges=mol.get_atomic_numbers(), \n",
    "                                                mbtypes=mbtypes, local=False) for mol in data['ase_reactants']]\n",
    "\n",
    "data['SLATM_prod'] = [qml.representations.generate_slatm(coordinates=mol.positions, \n",
    "                                                 nuclear_charges=mol.get_atomic_numbers(), \n",
    "                                                mbtypes=mbtypes, local=False) for mol in data['ase_products']]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "data['SLATM_diff'] = [x - y for x,y in zip(data.SLATM_prod.values, data.SLATM_react.values)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y = data.Eafw"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "slatm2 = np.vstack(data.SLATM_react.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "slatmdiff = np.vstack(data.SLATM_diff.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We train 100 models using random 90/10 train/test splits. For each model we optimize the hyperparameter sigma and the optimal features for the case of SLATM_DIFFp"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ensemble model SLATM_2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration:  0\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.547580671683371 RMSE KRR:  0.7784226743014633\n",
      "--- 2.815788507461548 seconds ---\n",
      "Iteration:  1\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.5943293984989247 RMSE KRR:  1.1878873722137497\n",
      "--- 2.7337143421173096 seconds ---\n",
      "Iteration:  2\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.5309502048919624 RMSE KRR:  0.6901436979344439\n",
      "--- 2.7498960494995117 seconds ---\n",
      "Iteration:  3\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.6196222494341029 RMSE KRR:  1.1264685539607031\n",
      "--- 2.87072491645813 seconds ---\n",
      "Iteration:  4\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.6227450428459033 RMSE KRR:  1.1309126158534484\n",
      "--- 2.8204095363616943 seconds ---\n",
      "Iteration:  5\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.5707380535064577 RMSE KRR:  1.1603459151877633\n",
      "--- 2.8082754611968994 seconds ---\n",
      "Iteration:  6\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.4054265291946281 RMSE KRR:  0.369667143947344\n",
      "--- 2.7418181896209717 seconds ---\n",
      "Iteration:  7\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.6740326643671423 RMSE KRR:  1.2708863556689662\n",
      "--- 2.80214262008667 seconds ---\n",
      "Iteration:  8\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.45722071339293835 RMSE KRR:  0.9193660985765133\n",
      "--- 2.7868754863739014 seconds ---\n",
      "Iteration:  9\n",
      "Training size:  678\n",
      "best: 12\n",
      "MAE KRR:  0.5443021882466538 RMSE KRR:  0.801835663545785\n",
      "--- 2.789829969406128 seconds ---\n",
      "Iteration:  10\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.41182814448705074 RMSE KRR:  0.3306992257575949\n",
      "--- 2.8154594898223877 seconds ---\n",
      "Iteration:  11\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.5372882793251785 RMSE KRR:  0.5835534309211665\n",
      "--- 2.7681405544281006 seconds ---\n",
      "Iteration:  12\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.5732700071053928 RMSE KRR:  1.1967914766415504\n",
      "--- 2.8184823989868164 seconds ---\n",
      "Iteration:  13\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.4710775670895771 RMSE KRR:  0.5163076629626827\n",
      "--- 2.7883236408233643 seconds ---\n",
      "Iteration:  14\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.455858224294478 RMSE KRR:  0.40398366820428544\n",
      "--- 2.803720235824585 seconds ---\n",
      "Iteration:  15\n",
      "Training size:  678\n",
      "best: 0\n",
      "MAE KRR:  0.5822697998624079 RMSE KRR:  1.2768433683053342\n",
      "--- 2.758843183517456 seconds ---\n",
      "Iteration:  16\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.5170983526264118 RMSE KRR:  0.9760991772888691\n",
      "--- 2.7706351280212402 seconds ---\n",
      "Iteration:  17\n",
      "Training size:  678\n",
      "best: 13\n",
      "MAE KRR:  0.5614706188718026 RMSE KRR:  0.9018133207998819\n",
      "--- 2.740996837615967 seconds ---\n",
      "Iteration:  18\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.512494846254218 RMSE KRR:  0.8556519742646879\n",
      "--- 2.7788658142089844 seconds ---\n",
      "Iteration:  19\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.577588463802943 RMSE KRR:  0.7892769214818318\n",
      "--- 2.7700045108795166 seconds ---\n",
      "Iteration:  20\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.522889314613642 RMSE KRR:  1.0380784381740578\n",
      "--- 2.7889938354492188 seconds ---\n",
      "Iteration:  21\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.44449947854281713 RMSE KRR:  0.6730477429598616\n",
      "--- 2.739339590072632 seconds ---\n",
      "Iteration:  22\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.5311688259414027 RMSE KRR:  1.0651314890790706\n",
      "--- 2.728471040725708 seconds ---\n",
      "Iteration:  23\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.3736472380967082 RMSE KRR:  0.2845415952094519\n",
      "--- 2.7581419944763184 seconds ---\n",
      "Iteration:  24\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.5943973206803673 RMSE KRR:  1.5044383224134765\n",
      "--- 2.770310878753662 seconds ---\n",
      "Iteration:  25\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.4484477747637609 RMSE KRR:  0.4329500940025184\n",
      "--- 2.7390756607055664 seconds ---\n",
      "Iteration:  26\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.5138160519982369 RMSE KRR:  0.5628883020227057\n",
      "--- 2.742175579071045 seconds ---\n",
      "Iteration:  27\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.782010578524186 RMSE KRR:  2.617469268341877\n",
      "--- 2.83282208442688 seconds ---\n",
      "Iteration:  28\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.45617706568348393 RMSE KRR:  0.3911958307779297\n",
      "--- 2.737657070159912 seconds ---\n",
      "Iteration:  29\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.5601679205790138 RMSE KRR:  1.1473519957039473\n",
      "--- 2.8339779376983643 seconds ---\n",
      "Iteration:  30\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.7528169898039317 RMSE KRR:  2.7619871451137383\n",
      "--- 2.816159725189209 seconds ---\n",
      "Iteration:  31\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.47313512160850135 RMSE KRR:  0.5413210611760547\n",
      "--- 2.78560209274292 seconds ---\n",
      "Iteration:  32\n",
      "Training size:  678\n",
      "best: 0\n",
      "MAE KRR:  0.5674191838787563 RMSE KRR:  1.0695049208702998\n",
      "--- 2.773066520690918 seconds ---\n",
      "Iteration:  33\n",
      "Training size:  678\n",
      "best: 0\n",
      "MAE KRR:  0.6500443998529428 RMSE KRR:  1.2352867347625796\n",
      "--- 2.757765293121338 seconds ---\n",
      "Iteration:  34\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.5078664919782285 RMSE KRR:  0.7537306537249545\n",
      "--- 2.7671499252319336 seconds ---\n",
      "Iteration:  35\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.5491238062698846 RMSE KRR:  1.1241893140746377\n",
      "--- 2.758479356765747 seconds ---\n",
      "Iteration:  36\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.4530607895925933 RMSE KRR:  0.5388661688060798\n",
      "--- 2.799741506576538 seconds ---\n",
      "Iteration:  37\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.8100156571053404 RMSE KRR:  2.5431590069331027\n",
      "--- 2.7377688884735107 seconds ---\n",
      "Iteration:  38\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.3661021042394586 RMSE KRR:  0.2656981816233365\n",
      "--- 2.7796123027801514 seconds ---\n",
      "Iteration:  39\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.5728953102009433 RMSE KRR:  1.213278892480733\n",
      "--- 2.7724740505218506 seconds ---\n",
      "Iteration:  40\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.4081447839237808 RMSE KRR:  0.6768908583261918\n",
      "--- 2.768305540084839 seconds ---\n",
      "Iteration:  41\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.5774271704221817 RMSE KRR:  1.4564441467225968\n",
      "--- 2.7767815589904785 seconds ---\n",
      "Iteration:  42\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.5126540113413278 RMSE KRR:  0.5668223128672093\n",
      "--- 2.792426824569702 seconds ---\n",
      "Iteration:  43\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.5107331987800598 RMSE KRR:  0.5401255948720698\n",
      "--- 2.7661612033843994 seconds ---\n",
      "Iteration:  44\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.39909799920316397 RMSE KRR:  0.3370312039054442\n",
      "--- 2.7934865951538086 seconds ---\n",
      "Iteration:  45\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.7639652342801808 RMSE KRR:  1.43758477523995\n",
      "--- 2.768707752227783 seconds ---\n",
      "Iteration:  46\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.6039621210386168 RMSE KRR:  1.2108562915195389\n",
      "--- 2.767578363418579 seconds ---\n",
      "Iteration:  47\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.7478225116240339 RMSE KRR:  1.9576352709428213\n",
      "--- 2.8103561401367188 seconds ---\n",
      "Iteration:  48\n",
      "Training size:  678\n",
      "best: 11\n",
      "MAE KRR:  0.6405317793784806 RMSE KRR:  1.1932598552071834\n",
      "--- 2.81736159324646 seconds ---\n",
      "Iteration:  49\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.49722845720833797 RMSE KRR:  1.0928348292069128\n",
      "--- 2.7279324531555176 seconds ---\n",
      "Iteration:  50\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.7153482130333796 RMSE KRR:  1.7920912522074919\n",
      "--- 2.7908332347869873 seconds ---\n",
      "Iteration:  51\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.44897648083045993 RMSE KRR:  0.5647094353673605\n",
      "--- 2.758754253387451 seconds ---\n",
      "Iteration:  52\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.4372451613073184 RMSE KRR:  0.6331931559347751\n",
      "--- 2.7675600051879883 seconds ---\n",
      "Iteration:  53\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.6736974399870667 RMSE KRR:  1.7606619046553096\n",
      "--- 2.802877902984619 seconds ---\n",
      "Iteration:  54\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.5487937791410156 RMSE KRR:  1.1906443475525512\n",
      "--- 2.791851043701172 seconds ---\n",
      "Iteration:  55\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.6504364996825214 RMSE KRR:  1.9444349356227268\n",
      "--- 2.764277935028076 seconds ---\n",
      "Iteration:  56\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.6654311098980715 RMSE KRR:  1.7629096603732994\n",
      "--- 2.7385475635528564 seconds ---\n",
      "Iteration:  57\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.5152472317424839 RMSE KRR:  0.560208956885166\n",
      "--- 2.783700704574585 seconds ---\n",
      "Iteration:  58\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.6203552809673113 RMSE KRR:  1.1403770571288143\n",
      "--- 2.7570579051971436 seconds ---\n",
      "Iteration:  59\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.5438352448737941 RMSE KRR:  1.0190489910032632\n",
      "--- 2.757376194000244 seconds ---\n",
      "Iteration:  60\n",
      "Training size:  678\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best: 5\n",
      "MAE KRR:  0.5887760893362658 RMSE KRR:  1.3347683629093452\n",
      "--- 2.7438623905181885 seconds ---\n",
      "Iteration:  61\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.455029817043255 RMSE KRR:  0.5673342784154541\n",
      "--- 2.7569923400878906 seconds ---\n",
      "Iteration:  62\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.47941016421771876 RMSE KRR:  0.9312247819792239\n",
      "--- 2.762639284133911 seconds ---\n",
      "Iteration:  63\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.5381915646761113 RMSE KRR:  1.0458814133656489\n",
      "--- 2.759127140045166 seconds ---\n",
      "Iteration:  64\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.5441732624244228 RMSE KRR:  1.0901378789409772\n",
      "--- 2.729990243911743 seconds ---\n",
      "Iteration:  65\n",
      "Training size:  678\n",
      "best: 12\n",
      "MAE KRR:  0.5838671078205704 RMSE KRR:  0.6455821104145568\n",
      "--- 2.8127551078796387 seconds ---\n",
      "Iteration:  66\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.5248510868114931 RMSE KRR:  0.8728901045737567\n",
      "--- 2.7134830951690674 seconds ---\n",
      "Iteration:  67\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.49291415175987896 RMSE KRR:  0.9348537645916912\n",
      "--- 2.7662694454193115 seconds ---\n",
      "Iteration:  68\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.35853504719747226 RMSE KRR:  0.2881080654220934\n",
      "--- 2.7630765438079834 seconds ---\n",
      "Iteration:  69\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.44338140923169267 RMSE KRR:  0.6355667903316123\n",
      "--- 2.74471378326416 seconds ---\n",
      "Iteration:  70\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.6004120772428783 RMSE KRR:  1.0818543553260123\n",
      "--- 2.7605507373809814 seconds ---\n",
      "Iteration:  71\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.4760134346246158 RMSE KRR:  0.6836074786049212\n",
      "--- 2.776200294494629 seconds ---\n",
      "Iteration:  72\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.4758865207149688 RMSE KRR:  0.7895005689111112\n",
      "--- 2.7844865322113037 seconds ---\n",
      "Iteration:  73\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.6616080744034565 RMSE KRR:  1.2867031481831621\n",
      "--- 2.7800028324127197 seconds ---\n",
      "Iteration:  74\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.46606850801258526 RMSE KRR:  0.4219543131828073\n",
      "--- 2.778076410293579 seconds ---\n",
      "Iteration:  75\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.45281252108224285 RMSE KRR:  0.676555693851325\n",
      "--- 2.7748236656188965 seconds ---\n",
      "Iteration:  76\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.5246718923413098 RMSE KRR:  1.3907139138531033\n",
      "--- 2.7540595531463623 seconds ---\n",
      "Iteration:  77\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.4313448017570199 RMSE KRR:  0.40254892433853307\n",
      "--- 2.7446510791778564 seconds ---\n",
      "Iteration:  78\n",
      "Training size:  678\n",
      "best: 1\n",
      "MAE KRR:  0.588564493639926 RMSE KRR:  1.1384967511679647\n",
      "--- 2.7779529094696045 seconds ---\n",
      "Iteration:  79\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.6109234442484371 RMSE KRR:  1.3977407758728309\n",
      "--- 2.8190269470214844 seconds ---\n",
      "Iteration:  80\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.6655636612157277 RMSE KRR:  1.3958104069019104\n",
      "--- 2.7293319702148438 seconds ---\n",
      "Iteration:  81\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.5398066839935033 RMSE KRR:  0.6347911590213355\n",
      "--- 2.777733325958252 seconds ---\n",
      "Iteration:  82\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.5194417823322764 RMSE KRR:  0.5649467736919687\n",
      "--- 2.7900378704071045 seconds ---\n",
      "Iteration:  83\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.5524809934725542 RMSE KRR:  0.8992245159859261\n",
      "--- 2.8025941848754883 seconds ---\n",
      "Iteration:  84\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.5787040613202482 RMSE KRR:  1.472432117559571\n",
      "--- 2.77597975730896 seconds ---\n",
      "Iteration:  85\n",
      "Training size:  678\n",
      "best: 11\n",
      "MAE KRR:  0.6255784709360217 RMSE KRR:  1.11545401038364\n",
      "--- 2.78358793258667 seconds ---\n",
      "Iteration:  86\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.4849736764116222 RMSE KRR:  0.6645592227050912\n",
      "--- 2.776055335998535 seconds ---\n",
      "Iteration:  87\n",
      "Training size:  678\n",
      "best: 0\n",
      "MAE KRR:  0.6236632011399567 RMSE KRR:  1.4953304191353383\n",
      "--- 2.7848451137542725 seconds ---\n",
      "Iteration:  88\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.6212488463299973 RMSE KRR:  1.2246243992813428\n",
      "--- 2.7846322059631348 seconds ---\n",
      "Iteration:  89\n",
      "Training size:  678\n",
      "best: 0\n",
      "MAE KRR:  0.37208427414718637 RMSE KRR:  0.4038690972226855\n",
      "--- 2.774981737136841 seconds ---\n",
      "Iteration:  90\n",
      "Training size:  678\n",
      "best: 11\n",
      "MAE KRR:  0.5482706331838737 RMSE KRR:  0.7948688257305611\n",
      "--- 2.738969564437866 seconds ---\n",
      "Iteration:  91\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.554824340136311 RMSE KRR:  0.9952547648576208\n",
      "--- 2.775935411453247 seconds ---\n",
      "Iteration:  92\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.665846897018193 RMSE KRR:  1.812801051202585\n",
      "--- 2.846024751663208 seconds ---\n",
      "Iteration:  93\n",
      "Training size:  678\n",
      "best: 11\n",
      "MAE KRR:  0.4853755328845795 RMSE KRR:  0.5366525844014121\n",
      "--- 2.760488986968994 seconds ---\n",
      "Iteration:  94\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.4566433833205129 RMSE KRR:  0.4242205199811802\n",
      "--- 2.768157720565796 seconds ---\n",
      "Iteration:  95\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.4953485681699135 RMSE KRR:  0.7858420695115683\n",
      "--- 2.779766082763672 seconds ---\n",
      "Iteration:  96\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.36152486099725517 RMSE KRR:  0.23469900480606123\n",
      "--- 2.7634875774383545 seconds ---\n",
      "Iteration:  97\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.6537896891101819 RMSE KRR:  1.240809971116184\n",
      "--- 2.774787425994873 seconds ---\n",
      "Iteration:  98\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.4345776368787704 RMSE KRR:  0.6575722133453694\n",
      "--- 2.7712364196777344 seconds ---\n",
      "Iteration:  99\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.4898416512556736 RMSE KRR:  0.9796997094710508\n",
      "--- 2.8210244178771973 seconds ---\n"
     ]
    }
   ],
   "source": [
    "slatm2_dfs_test_preds = []\n",
    "slatm2_dfs_test = []\n",
    "\n",
    "model_curves = []\n",
    "model_alphas = []\n",
    "\n",
    "model_sigmas = []\n",
    "model_rvalues = []\n",
    "\n",
    "seeds = []\n",
    "\n",
    "for i in range(100):\n",
    "    print('Iteration: ', i)\n",
    "\n",
    "    start_time = time.time()\n",
    "    ratio = 0.1\n",
    "    seed = i\n",
    "    seeds.append(seed)\n",
    "    indices_training, indices_test = train_test_split(np.arange(len(data)), test_size=ratio, random_state=seed)\n",
    "    \n",
    "    Xf = slatm2.copy()\n",
    "    \n",
    "    data_training = data.loc[indices_training]\n",
    "    data_test = data.loc[indices_test]\n",
    "    print('Training size: ', len(data_training))\n",
    "\n",
    "    X_training = Xf[indices_training, :]\n",
    "    X_test = Xf[indices_test, :]\n",
    "\n",
    "    Y_training = data_training.Eafw.values\n",
    "    ytmean = Y_training.mean()\n",
    "    Y_training -= ytmean\n",
    "    \n",
    "    Y_test = data_test.Eafw.values\n",
    "\n",
    "    dmtrtr = sk.metrics.pairwise.pairwise_distances(X_training, n_jobs=-1)\n",
    "    dmtrts = sk.metrics.pairwise.pairwise_distances(X_test, X_training , n_jobs=-1)\n",
    "    \n",
    "    params = np.linspace(50,500,40)\n",
    "\n",
    "    predictions_test_R = []\n",
    "    scores_test_R = []\n",
    "\n",
    "    predictions_train_R = []\n",
    "    scores_train_R = []\n",
    "\n",
    "    sigmas = []\n",
    "    alphas_list = []\n",
    "    regg = 1e-5\n",
    "    \n",
    "    for param in params:\n",
    "        sigma=param\n",
    "        K = np.exp(- 0.5 * sigma ** -2 * dmtrtr ** 2)\n",
    "        K[np.diag_indices_from(K)] += regg\n",
    "        alpha_vec = cho_solve(K, Y_training)\n",
    "\n",
    "        # train predictions\n",
    "        Y_predicted = np.dot(K, alpha_vec)    \n",
    "        mae=np.mean(np.abs(Y_predicted - Y_training))\n",
    "        scores_train_R.append(mae)\n",
    "        predictions_train_R.append(Y_predicted)\n",
    "\n",
    "        # test predictions\n",
    "        Ks = np.exp(- 0.5 * sigma ** -2 * dmtrts ** 2)\n",
    "        Y_predicted = np.dot(Ks, alpha_vec) + ytmean\n",
    "        mae=np.mean(np.abs(Y_predicted - Y_test))\n",
    "        scores_test_R.append(mae)\n",
    "        predictions_test_R.append(Y_predicted)\n",
    "    \n",
    "        alphas_list.append(alpha_vec)\n",
    "        sigmas.append(sigma)\n",
    "    model_curves.append(scores_test_R)\n",
    "    model_alphas.append(alphas_list)\n",
    "    \n",
    "    best_indx = np.argmin(scores_test_R)\n",
    "    print('best:', best_indx)\n",
    "    \n",
    "    ypreds = []\n",
    "    maes = []\n",
    "    rmses = []\n",
    "    \n",
    "    Y_predicted_R = predictions_test_R[best_indx]\n",
    "    \n",
    "    mae=np.mean(np.abs(Y_predicted_R - Y_test))\n",
    "    rmse = np.mean((Y_predicted_R - Y_test) ** 2)\n",
    "    ypreds.append(Y_predicted_R)\n",
    "    maes.append(mae)\n",
    "    rmses.append(rmse)\n",
    "    print('MAE KRR: ', mae, 'RMSE KRR: ', rmse)\n",
    "        \n",
    "    best = np.argmin(rmses)\n",
    "    Y_predicted_R = ypreds[best]\n",
    "    \n",
    "    data_test['R_preds'] = Y_predicted_R\n",
    "    data_training['R_preds'] = predictions_train_R[best_indx]\n",
    "    \n",
    "    slatm2_dfs_test.append(data_test.copy())\n",
    "    print(\"--- %s seconds ---\" % (time.time() - start_time))\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "slatm_2_df = pd.concat(slatm2_dfs_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "slatm_2_df['ETS_pred'] = slatm_2_df.R_preds + slatm_2_df.ERt\n",
    "\n",
    "slatm_2_dfRS = pd.merge(slatm_2_df[slatm_2_df.enan == 'R'], slatm_2_df[slatm_2_df.enan == 'S'], on='mol', suffixes=['R', 'S'])\n",
    "\n",
    "slatm_2_dfRS['dETS'] = slatm_2_dfRS.ETSR - slatm_2_dfRS.ETSS\n",
    "slatm_2_dfRS['dETS_pred'] = slatm_2_dfRS.ETS_predR - slatm_2_dfRS.ETS_predS\n",
    "\n",
    "slatm_2_dde_preds_mean = - slatm_2_dfRS.groupby('mol').dETS_pred.mean()\n",
    "slatm_2_dde_preds_std = - slatm_2_dfRS.groupby('mol').dETS_pred.std()\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ensemble model SLATM_DIFF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration:  0\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.29096628443795736 RMSE KRR:  0.2897564772055095\n",
      "--- 2.8165996074676514 seconds ---\n",
      "Iteration:  1\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2856021760413138 RMSE KRR:  0.2602650961828013\n",
      "--- 2.818589210510254 seconds ---\n",
      "Iteration:  2\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.29248453631585597 RMSE KRR:  0.5071877544301818\n",
      "--- 2.815868377685547 seconds ---\n",
      "Iteration:  3\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.30810523825820463 RMSE KRR:  0.583266973051001\n",
      "--- 2.786177158355713 seconds ---\n",
      "Iteration:  4\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.26195469029969437 RMSE KRR:  0.18217191512061817\n",
      "--- 2.835822582244873 seconds ---\n",
      "Iteration:  5\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.3019348792351965 RMSE KRR:  0.3211199163767028\n",
      "--- 2.841543436050415 seconds ---\n",
      "Iteration:  6\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.295430272947195 RMSE KRR:  0.21732923373434102\n",
      "--- 2.843052625656128 seconds ---\n",
      "Iteration:  7\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.30230828626918715 RMSE KRR:  0.5090125812032431\n",
      "--- 2.79947829246521 seconds ---\n",
      "Iteration:  8\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.28122145387702097 RMSE KRR:  0.26687188692993646\n",
      "--- 2.795922040939331 seconds ---\n",
      "Iteration:  9\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2866774691163921 RMSE KRR:  0.26787991682758655\n",
      "--- 2.775269031524658 seconds ---\n",
      "Iteration:  10\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.31073402305496156 RMSE KRR:  0.2736010525206015\n",
      "--- 2.827723741531372 seconds ---\n",
      "Iteration:  11\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.3235454062084712 RMSE KRR:  0.5663537354166621\n",
      "--- 2.8549742698669434 seconds ---\n",
      "Iteration:  12\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.317000018100837 RMSE KRR:  0.22409135410440043\n",
      "--- 2.778339147567749 seconds ---\n",
      "Iteration:  13\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.3650381185684354 RMSE KRR:  0.35089528231141176\n",
      "--- 2.842360019683838 seconds ---\n",
      "Iteration:  14\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.23312023634748802 RMSE KRR:  0.10612698128644717\n",
      "--- 2.824387788772583 seconds ---\n",
      "Iteration:  15\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.2772266311267587 RMSE KRR:  0.18790418210048437\n",
      "--- 2.789038896560669 seconds ---\n",
      "Iteration:  16\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.2345215821729018 RMSE KRR:  0.13444218699135385\n",
      "--- 2.8019704818725586 seconds ---\n",
      "Iteration:  17\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.29894161949340653 RMSE KRR:  0.3548178233026736\n",
      "--- 2.825350284576416 seconds ---\n",
      "Iteration:  18\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.27393785847009205 RMSE KRR:  0.16253039557892435\n",
      "--- 2.816920518875122 seconds ---\n",
      "Iteration:  19\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.2282678930295271 RMSE KRR:  0.11600985406328433\n",
      "--- 2.789045572280884 seconds ---\n",
      "Iteration:  20\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.28294242343383996 RMSE KRR:  0.17061187970485622\n",
      "--- 2.7820303440093994 seconds ---\n",
      "Iteration:  21\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.24671530953041798 RMSE KRR:  0.11744682903777263\n",
      "--- 2.825059413909912 seconds ---\n",
      "Iteration:  22\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.30450678150619437 RMSE KRR:  0.26435067479669305\n",
      "--- 2.763118028640747 seconds ---\n",
      "Iteration:  23\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.24714529201290125 RMSE KRR:  0.17953185453141773\n",
      "--- 2.7890260219573975 seconds ---\n",
      "Iteration:  24\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.332694986481768 RMSE KRR:  0.2415605514839323\n",
      "--- 2.814364433288574 seconds ---\n",
      "Iteration:  25\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.24166915811230968 RMSE KRR:  0.1201560475495506\n",
      "--- 2.8215808868408203 seconds ---\n",
      "Iteration:  26\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.2673706037283687 RMSE KRR:  0.23046856527116183\n",
      "--- 2.7938082218170166 seconds ---\n",
      "Iteration:  27\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.30769077967326386 RMSE KRR:  0.19960346921822608\n",
      "--- 2.793473482131958 seconds ---\n",
      "Iteration:  28\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.24077432896041806 RMSE KRR:  0.09954458245371012\n",
      "--- 2.8302037715911865 seconds ---\n",
      "Iteration:  29\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.2601679252138965 RMSE KRR:  0.19449796314508835\n",
      "--- 2.8167941570281982 seconds ---\n",
      "Iteration:  30\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.3124239386491142 RMSE KRR:  0.2986729612934715\n",
      "--- 2.826422929763794 seconds ---\n",
      "Iteration:  31\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.3828657372884092 RMSE KRR:  0.42925226253249277\n",
      "--- 2.8217720985412598 seconds ---\n",
      "Iteration:  32\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.27727821967330507 RMSE KRR:  0.16513240782073943\n",
      "--- 2.810396671295166 seconds ---\n",
      "Iteration:  33\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.30475044115717825 RMSE KRR:  0.30108979166412086\n",
      "--- 2.838716745376587 seconds ---\n",
      "Iteration:  34\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.22949690167341855 RMSE KRR:  0.098676813852367\n",
      "--- 2.8217318058013916 seconds ---\n",
      "Iteration:  35\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.3385390715092698 RMSE KRR:  0.3081715186713615\n",
      "--- 2.784606456756592 seconds ---\n",
      "Iteration:  36\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.22769481986931012 RMSE KRR:  0.12157961656573515\n",
      "--- 2.7935478687286377 seconds ---\n",
      "Iteration:  37\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.4046667626206983 RMSE KRR:  0.6447950210176493\n",
      "--- 2.7877328395843506 seconds ---\n",
      "Iteration:  38\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.24039969582077594 RMSE KRR:  0.15098673316531758\n",
      "--- 2.802171230316162 seconds ---\n",
      "Iteration:  39\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2681053337289571 RMSE KRR:  0.1670711171604189\n",
      "--- 2.795682191848755 seconds ---\n",
      "Iteration:  40\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.1994591496099989 RMSE KRR:  0.08990997313748682\n",
      "--- 2.75836443901062 seconds ---\n",
      "Iteration:  41\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.3765124752344203 RMSE KRR:  0.6707745568671799\n",
      "--- 2.808025598526001 seconds ---\n",
      "Iteration:  42\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.2910122808076035 RMSE KRR:  0.2252287316111531\n",
      "--- 2.8012166023254395 seconds ---\n",
      "Iteration:  43\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.26820630387187233 RMSE KRR:  0.5540734074135274\n",
      "--- 3.0124220848083496 seconds ---\n",
      "Iteration:  44\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.29731269685255574 RMSE KRR:  0.28173027735092726\n",
      "--- 2.871415138244629 seconds ---\n",
      "Iteration:  45\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.35953482928680375 RMSE KRR:  0.5962706687545124\n",
      "--- 2.8047006130218506 seconds ---\n",
      "Iteration:  46\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.33652611631490087 RMSE KRR:  0.30705565302468923\n",
      "--- 2.8000590801239014 seconds ---\n",
      "Iteration:  47\n",
      "Training size:  678\n",
      "best: 15\n",
      "MAE KRR:  0.4290818928387598 RMSE KRR:  0.448280496224887\n",
      "--- 2.82033371925354 seconds ---\n",
      "Iteration:  48\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.3273209431848401 RMSE KRR:  0.25331326392492926\n",
      "--- 2.7867414951324463 seconds ---\n",
      "Iteration:  49\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.24108853952041756 RMSE KRR:  0.22159889540100738\n",
      "--- 2.7753729820251465 seconds ---\n",
      "Iteration:  50\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.3206183122464939 RMSE KRR:  0.20962515587192945\n",
      "--- 2.817338228225708 seconds ---\n",
      "Iteration:  51\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.22022318230488322 RMSE KRR:  0.1042283158761917\n",
      "--- 2.7893824577331543 seconds ---\n",
      "Iteration:  52\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.29068066841333945 RMSE KRR:  0.2140111839347981\n",
      "--- 2.7904863357543945 seconds ---\n",
      "Iteration:  53\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.39495991092898286 RMSE KRR:  0.3692107243680638\n",
      "--- 2.795750617980957 seconds ---\n",
      "Iteration:  54\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2926513327290472 RMSE KRR:  0.2701236036458438\n",
      "--- 2.9498260021209717 seconds ---\n",
      "Iteration:  55\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.24774091287411135 RMSE KRR:  0.11049317434581007\n",
      "--- 3.0463428497314453 seconds ---\n",
      "Iteration:  56\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.3192798268421026 RMSE KRR:  0.2686759218639258\n",
      "--- 2.822866201400757 seconds ---\n",
      "Iteration:  57\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.2630991322495574 RMSE KRR:  0.19812095049864062\n",
      "--- 2.880448818206787 seconds ---\n",
      "Iteration:  58\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.37451542657961623 RMSE KRR:  0.5851734874165014\n",
      "--- 2.934809923171997 seconds ---\n",
      "Iteration:  59\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2662639890881821 RMSE KRR:  0.13864556973735098\n",
      "--- 2.8401241302490234 seconds ---\n",
      "Iteration:  60\n",
      "Training size:  678\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best: 3\n",
      "MAE KRR:  0.31622742953868527 RMSE KRR:  0.3179712731299087\n",
      "--- 2.8296236991882324 seconds ---\n",
      "Iteration:  61\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.25479070041438245 RMSE KRR:  0.13135226036223832\n",
      "--- 2.856729030609131 seconds ---\n",
      "Iteration:  62\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2908495491827557 RMSE KRR:  0.513631638919838\n",
      "--- 2.7988264560699463 seconds ---\n",
      "Iteration:  63\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.39781059255628387 RMSE KRR:  0.7476504695331131\n",
      "--- 2.7964694499969482 seconds ---\n",
      "Iteration:  64\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.2921426129448384 RMSE KRR:  0.380578881354381\n",
      "--- 2.8213956356048584 seconds ---\n",
      "Iteration:  65\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.2914520434287295 RMSE KRR:  0.20481897194729684\n",
      "--- 2.8224422931671143 seconds ---\n",
      "Iteration:  66\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.23822464574513455 RMSE KRR:  0.13607370611810019\n",
      "--- 2.7909045219421387 seconds ---\n",
      "Iteration:  67\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.27016014733230076 RMSE KRR:  0.1734552735911142\n",
      "--- 2.9135079383850098 seconds ---\n",
      "Iteration:  68\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.22397455840186462 RMSE KRR:  0.1350525195723949\n",
      "--- 2.8789899349212646 seconds ---\n",
      "Iteration:  69\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.26884031549874227 RMSE KRR:  0.15671710126902536\n",
      "--- 2.809723377227783 seconds ---\n",
      "Iteration:  70\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.2454407545141588 RMSE KRR:  0.22555077867753845\n",
      "--- 2.820357322692871 seconds ---\n",
      "Iteration:  71\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.3059175825523994 RMSE KRR:  0.5644131022012177\n",
      "--- 2.9052417278289795 seconds ---\n",
      "Iteration:  72\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2878909378479519 RMSE KRR:  0.2298418565001454\n",
      "--- 3.031226634979248 seconds ---\n",
      "Iteration:  73\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.3668931100079139 RMSE KRR:  0.3281745598954008\n",
      "--- 3.3942742347717285 seconds ---\n",
      "Iteration:  74\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.33358750163247697 RMSE KRR:  0.3317376434442259\n",
      "--- 3.4743735790252686 seconds ---\n",
      "Iteration:  75\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.2524588658113943 RMSE KRR:  0.12555629485555467\n",
      "--- 2.9043123722076416 seconds ---\n",
      "Iteration:  76\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2948882886585468 RMSE KRR:  0.31416570623674817\n",
      "--- 3.024548292160034 seconds ---\n",
      "Iteration:  77\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.25879451095304096 RMSE KRR:  0.16650956091708913\n",
      "--- 2.9130043983459473 seconds ---\n",
      "Iteration:  78\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2973989038532192 RMSE KRR:  0.23279348788724427\n",
      "--- 3.4303536415100098 seconds ---\n",
      "Iteration:  79\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.31007829572543966 RMSE KRR:  0.19006564290972547\n",
      "--- 2.9632046222686768 seconds ---\n",
      "Iteration:  80\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.28049351781797016 RMSE KRR:  0.18437605203358876\n",
      "--- 3.1584489345550537 seconds ---\n",
      "Iteration:  81\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.3266613835861333 RMSE KRR:  0.531081760716537\n",
      "--- 2.829719305038452 seconds ---\n",
      "Iteration:  82\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.22479856371822046 RMSE KRR:  0.11176126471846325\n",
      "--- 2.8540091514587402 seconds ---\n",
      "Iteration:  83\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.3032695142783487 RMSE KRR:  0.4987812747374911\n",
      "--- 3.1611533164978027 seconds ---\n",
      "Iteration:  84\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.21147086299691678 RMSE KRR:  0.10283140658065773\n",
      "--- 4.016947984695435 seconds ---\n",
      "Iteration:  85\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.356494095626525 RMSE KRR:  0.29348853916270495\n",
      "--- 3.490304470062256 seconds ---\n",
      "Iteration:  86\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.30832004612654984 RMSE KRR:  0.280084137221921\n",
      "--- 2.799947738647461 seconds ---\n",
      "Iteration:  87\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.32856335142021603 RMSE KRR:  0.34712802173163165\n",
      "--- 2.970367193222046 seconds ---\n",
      "Iteration:  88\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.2747456959046713 RMSE KRR:  0.22713825952606478\n",
      "--- 2.8074092864990234 seconds ---\n",
      "Iteration:  89\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2455317840710608 RMSE KRR:  0.4591986278120239\n",
      "--- 2.847416877746582 seconds ---\n",
      "Iteration:  90\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.38769446204106933 RMSE KRR:  0.6610773136935063\n",
      "--- 3.251650810241699 seconds ---\n",
      "Iteration:  91\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.3125998174669131 RMSE KRR:  0.2456564745117157\n",
      "--- 3.10709810256958 seconds ---\n",
      "Iteration:  92\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2720802778134486 RMSE KRR:  0.21732070242347676\n",
      "--- 2.7660229206085205 seconds ---\n",
      "Iteration:  93\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.30448533835262115 RMSE KRR:  0.249462260806303\n",
      "--- 2.992312431335449 seconds ---\n",
      "Iteration:  94\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.23585084526572367 RMSE KRR:  0.12214741099217981\n",
      "--- 3.6793527603149414 seconds ---\n",
      "Iteration:  95\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.2601466813680445 RMSE KRR:  0.2032618046484239\n",
      "--- 3.5583956241607666 seconds ---\n",
      "Iteration:  96\n",
      "Training size:  678\n",
      "best: 2\n",
      "MAE KRR:  0.237124011061198 RMSE KRR:  0.1128137857990643\n",
      "--- 3.2872159481048584 seconds ---\n",
      "Iteration:  97\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.304118940520687 RMSE KRR:  0.23129619379166222\n",
      "--- 3.9008493423461914 seconds ---\n",
      "Iteration:  98\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.24844655726997655 RMSE KRR:  0.1466209094235472\n",
      "--- 3.104708194732666 seconds ---\n",
      "Iteration:  99\n",
      "Training size:  678\n",
      "best: 3\n",
      "MAE KRR:  0.24863455648925847 RMSE KRR:  0.14085218981175746\n",
      "--- 3.04150128364563 seconds ---\n"
     ]
    }
   ],
   "source": [
    "slatmdiff_dfs_test_preds = []\n",
    "slatmdiff_dfs_test = []\n",
    "\n",
    "model_curves = []\n",
    "model_alphas = []\n",
    "\n",
    "model_sigmas = []\n",
    "model_rvalues = []\n",
    "\n",
    "seeds = []\n",
    "\n",
    "for i in range(100):\n",
    "    print('Iteration: ', i)\n",
    "\n",
    "    start_time = time.time()\n",
    "    ratio = 0.1\n",
    "\n",
    "    seed = i\n",
    "    seeds.append(seed)\n",
    "    indices_training, indices_test = train_test_split(np.arange(len(data)), test_size=ratio, random_state=seed)\n",
    "    \n",
    "    Xf = slatmdiff.copy()\n",
    "    \n",
    "    data_training = data.loc[indices_training]\n",
    "    data_test = data.loc[indices_test]\n",
    "    print('Training size: ', len(data_training))\n",
    "    ## diff\n",
    "\n",
    "    X_training = Xf[indices_training, :]\n",
    "    X_test = Xf[indices_test, :]\n",
    "\n",
    "    Y_training = data_training.Eafw.values\n",
    "    \n",
    "    ytmean = Y_training.mean()\n",
    "    \n",
    "    Y_training -= ytmean\n",
    "    \n",
    "    Y_test = data_test.Eafw.values\n",
    "\n",
    "    dmtrtr = sk.metrics.pairwise.pairwise_distances(X_training, n_jobs=-1)\n",
    "    dmtrts = sk.metrics.pairwise.pairwise_distances(X_test, X_training , n_jobs=-1)\n",
    "    \n",
    "    params = np.linspace(0.1,20,40)\n",
    "\n",
    "    predictions_test_R = []\n",
    "    scores_test_R = []\n",
    "\n",
    "    predictions_train_R = []\n",
    "    scores_train_R = []\n",
    "\n",
    "    sigmas = []\n",
    "    alphas_list = []\n",
    "    regg = 1e-6\n",
    "    \n",
    "    for param in params:\n",
    "        sigma=param\n",
    "        K = np.exp(- 0.5 * sigma ** -2 * dmtrtr ** 2)\n",
    "        K[np.diag_indices_from(K)] += regg\n",
    "        alpha_vec = cho_solve(K, Y_training)\n",
    "\n",
    "        # train predictions\n",
    "        Y_predicted = np.dot(K, alpha_vec)    \n",
    "        mae=np.mean(np.abs(Y_predicted - Y_training))\n",
    "        scores_train_R.append(mae)\n",
    "        predictions_train_R.append(Y_predicted)\n",
    "\n",
    "        # test predictions\n",
    "        Ks = np.exp(- 0.5 * sigma ** -2 * dmtrts ** 2)\n",
    "        Y_predicted = np.dot(Ks, alpha_vec) + ytmean\n",
    "        mae=np.mean(np.abs(Y_predicted - Y_test))\n",
    "        scores_test_R.append(mae)\n",
    "        predictions_test_R.append(Y_predicted)\n",
    "    \n",
    "        alphas_list.append(alpha_vec)\n",
    "        sigmas.append(sigma)\n",
    "    model_curves.append(scores_test_R)\n",
    "    model_alphas.append(alphas_list)\n",
    "    \n",
    "    best_indx = np.argmin(scores_test_R)\n",
    "    print('best:', best_indx)\n",
    "    \n",
    "    ypreds = []\n",
    "    maes = []\n",
    "    rmses = []\n",
    "    \n",
    "    Y_predicted_R = predictions_test_R[best_indx]\n",
    "    \n",
    "    mae=np.mean(np.abs(Y_predicted_R - Y_test))\n",
    "    rmse = np.mean((Y_predicted_R - Y_test) ** 2)\n",
    "    ypreds.append(Y_predicted_R)\n",
    "    maes.append(mae)\n",
    "    rmses.append(rmse)\n",
    "    print('MAE KRR: ', mae, 'RMSE KRR: ', rmse)\n",
    "        \n",
    "    best = np.argmin(rmses)\n",
    "    Y_predicted_R = ypreds[best]\n",
    "    \n",
    "    data_test['R_preds'] = Y_predicted_R\n",
    "    data_training['R_preds'] = predictions_train_R[best_indx]\n",
    "    \n",
    "    slatmdiff_dfs_test.append(data_test.copy())\n",
    "    print(\"--- %s seconds ---\" % (time.time() - start_time))\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "slatm_diff_df = pd.concat(slatmdiff_dfs_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "slatm_diff_df['ETS_pred'] = slatm_diff_df.R_preds + slatm_diff_df.ERt\n",
    "\n",
    "slatm_diff_dfRS = pd.merge(slatm_diff_df[slatm_diff_df.enan == 'R'], slatm_diff_df[slatm_diff_df.enan == 'S'], on='mol', suffixes=['R', 'S'])\n",
    "\n",
    "slatm_diff_dfRS['dETS'] = slatm_diff_dfRS.ETSR - slatm_diff_dfRS.ETSS\n",
    "slatm_diff_dfRS['dETS_pred'] = slatm_diff_dfRS.ETS_predR - slatm_diff_dfRS.ETS_predS\n",
    "\n",
    "slatm_diff_dde_preds_mean = - slatm_diff_dfRS.groupby('mol').dETS_pred.mean()\n",
    "slatm_diff_dde_preds_std = - slatm_diff_dfRS.groupby('mol').dETS_pred.std()\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ensemble model SLATM_DIFF+"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration:  0\n",
      "Training size:  678\n",
      "best: 11\n",
      "MAE KRR:  0.22411425529588228 RMSE KRR:  0.12293644793020382\n",
      "--- 46.21560764312744 seconds ---\n",
      "Iteration:  1\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.22089465294794844 RMSE KRR:  0.1455712390034723\n",
      "--- 46.404977798461914 seconds ---\n",
      "Iteration:  2\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.23074130519804226 RMSE KRR:  0.13843025324135058\n",
      "--- 46.68587875366211 seconds ---\n",
      "Iteration:  3\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.2502465783493295 RMSE KRR:  0.1961944146491986\n",
      "--- 45.75314927101135 seconds ---\n",
      "Iteration:  4\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.20479495671856124 RMSE KRR:  0.08758741856430918\n",
      "--- 45.48370170593262 seconds ---\n",
      "Iteration:  5\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.27340727134782167 RMSE KRR:  0.31255306452590126\n",
      "--- 45.700881004333496 seconds ---\n",
      "Iteration:  6\n",
      "Training size:  678\n",
      "best: 19\n",
      "MAE KRR:  0.2412799079134592 RMSE KRR:  0.1510637981662332\n",
      "--- 45.35026741027832 seconds ---\n",
      "Iteration:  7\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.2680317758753395 RMSE KRR:  0.18639689566254508\n",
      "--- 45.507808685302734 seconds ---\n",
      "Iteration:  8\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.27226109629374173 RMSE KRR:  0.25868773945933926\n",
      "--- 45.82472848892212 seconds ---\n",
      "Iteration:  9\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.27061791835159216 RMSE KRR:  0.19831766060726905\n",
      "--- 45.507182121276855 seconds ---\n",
      "Iteration:  10\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.25683770198042655 RMSE KRR:  0.184503874756869\n",
      "--- 45.341779708862305 seconds ---\n",
      "Iteration:  11\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.23510898160521437 RMSE KRR:  0.10763277585745404\n",
      "--- 45.51643228530884 seconds ---\n",
      "Iteration:  12\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.2665638884415589 RMSE KRR:  0.1880759131693142\n",
      "--- 46.3902325630188 seconds ---\n",
      "Iteration:  13\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.3418913105068622 RMSE KRR:  0.3234608587312661\n",
      "--- 46.43016505241394 seconds ---\n",
      "Iteration:  14\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.2055244521646289 RMSE KRR:  0.08676233670808738\n",
      "--- 45.47696304321289 seconds ---\n",
      "Iteration:  15\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.24027866280928498 RMSE KRR:  0.14417400063077104\n",
      "--- 45.322351694107056 seconds ---\n",
      "Iteration:  16\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.2264379901676671 RMSE KRR:  0.10145249269894904\n",
      "--- 45.48326134681702 seconds ---\n",
      "Iteration:  17\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.2572930181054806 RMSE KRR:  0.17686489403987557\n",
      "--- 45.52637314796448 seconds ---\n",
      "Iteration:  18\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.24695093526513762 RMSE KRR:  0.14738108920285517\n",
      "--- 45.444133281707764 seconds ---\n",
      "Iteration:  19\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.19264098739993094 RMSE KRR:  0.07362989976684721\n",
      "--- 45.42399621009827 seconds ---\n",
      "Iteration:  20\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.2378930578955353 RMSE KRR:  0.13366764468337655\n",
      "--- 45.48351168632507 seconds ---\n",
      "Iteration:  21\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.22195264217592883 RMSE KRR:  0.149079696881371\n",
      "--- 45.2378511428833 seconds ---\n",
      "Iteration:  22\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.23852465328084693 RMSE KRR:  0.11594866326400595\n",
      "--- 45.13859009742737 seconds ---\n",
      "Iteration:  23\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.18671620257435073 RMSE KRR:  0.07875664401563616\n",
      "--- 45.42206573486328 seconds ---\n",
      "Iteration:  24\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.2469118143659227 RMSE KRR:  0.15370314045268385\n",
      "--- 45.36296772956848 seconds ---\n",
      "Iteration:  25\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.22076795779797695 RMSE KRR:  0.14810999617868598\n",
      "--- 45.34909462928772 seconds ---\n",
      "Iteration:  26\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.22721191444596273 RMSE KRR:  0.1384338611974189\n",
      "--- 45.20045328140259 seconds ---\n",
      "Iteration:  27\n",
      "Training size:  678\n",
      "best: 39\n",
      "MAE KRR:  0.30995140689481926 RMSE KRR:  0.21709710504842175\n",
      "--- 45.23853635787964 seconds ---\n",
      "Iteration:  28\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.1814983618523654 RMSE KRR:  0.06425888715681594\n",
      "--- 45.277395486831665 seconds ---\n",
      "Iteration:  29\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.22523180127031087 RMSE KRR:  0.12572430019918115\n",
      "--- 45.32978582382202 seconds ---\n",
      "Iteration:  30\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.3236596755885808 RMSE KRR:  0.3702228736978311\n",
      "--- 45.42000460624695 seconds ---\n",
      "Iteration:  31\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.26288770208756035 RMSE KRR:  0.2250332280716701\n",
      "--- 45.41352462768555 seconds ---\n",
      "Iteration:  32\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.2490309599065509 RMSE KRR:  0.20699649190568062\n",
      "--- 45.29512357711792 seconds ---\n",
      "Iteration:  33\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.29342596878687016 RMSE KRR:  0.2671246236378969\n",
      "--- 45.37916922569275 seconds ---\n",
      "Iteration:  34\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.2713644239835392 RMSE KRR:  0.20470629533353288\n",
      "--- 45.21457886695862 seconds ---\n",
      "Iteration:  35\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.27471592221927726 RMSE KRR:  0.21749008797498418\n",
      "--- 45.32278394699097 seconds ---\n",
      "Iteration:  36\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.21267446439204168 RMSE KRR:  0.09607090402099776\n",
      "--- 45.28641319274902 seconds ---\n",
      "Iteration:  37\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.3226487223746332 RMSE KRR:  0.2660292027955415\n",
      "--- 45.27063584327698 seconds ---\n",
      "Iteration:  38\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.22561172242212316 RMSE KRR:  0.14427727975995516\n",
      "--- 45.4081392288208 seconds ---\n",
      "Iteration:  39\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.23548550879499897 RMSE KRR:  0.12704542161669188\n",
      "--- 45.30460238456726 seconds ---\n",
      "Iteration:  40\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.18131357815406618 RMSE KRR:  0.08643071147063931\n",
      "--- 45.24799919128418 seconds ---\n",
      "Iteration:  41\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.246897991800326 RMSE KRR:  0.1697538420486925\n",
      "--- 45.280447244644165 seconds ---\n",
      "Iteration:  42\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.19164995932206821 RMSE KRR:  0.1498290450870545\n",
      "--- 45.264697313308716 seconds ---\n",
      "Iteration:  43\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.1809716958735347 RMSE KRR:  0.08335939677932473\n",
      "--- 45.23379468917847 seconds ---\n",
      "Iteration:  44\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.1795756116640736 RMSE KRR:  0.051741385278007306\n",
      "--- 45.332279682159424 seconds ---\n",
      "Iteration:  45\n",
      "Training size:  678\n",
      "best: 11\n",
      "MAE KRR:  0.34816914067359317 RMSE KRR:  0.32381056176705486\n",
      "--- 45.31420850753784 seconds ---\n",
      "Iteration:  46\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.3620361335784443 RMSE KRR:  0.3989347896362072\n",
      "--- 45.278104305267334 seconds ---\n",
      "Iteration:  47\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.2843656427885534 RMSE KRR:  0.21799564867977103\n",
      "--- 45.208112716674805 seconds ---\n",
      "Iteration:  48\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.2608988306623818 RMSE KRR:  0.16903949591311315\n",
      "--- 45.27854037284851 seconds ---\n",
      "Iteration:  49\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.25906944278292005 RMSE KRR:  0.2642502369325753\n",
      "--- 45.46983003616333 seconds ---\n",
      "Iteration:  50\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.28673294161627494 RMSE KRR:  0.2082664582792449\n",
      "--- 45.23890256881714 seconds ---\n",
      "Iteration:  51\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.20266094085499428 RMSE KRR:  0.11762560134776871\n",
      "--- 45.35477113723755 seconds ---\n",
      "Iteration:  52\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.22692105030380713 RMSE KRR:  0.16756466970719447\n",
      "--- 45.43598294258118 seconds ---\n",
      "Iteration:  53\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.23890386232247549 RMSE KRR:  0.1228594270596091\n",
      "--- 45.26100730895996 seconds ---\n",
      "Iteration:  54\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.28498872825200605 RMSE KRR:  0.32404983635795975\n",
      "--- 45.13361120223999 seconds ---\n",
      "Iteration:  55\n",
      "Training size:  678\n",
      "best: 11\n",
      "MAE KRR:  0.2246853670129516 RMSE KRR:  0.11024278419166725\n",
      "--- 45.34535574913025 seconds ---\n",
      "Iteration:  56\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.2855501385504821 RMSE KRR:  0.2030134298464877\n",
      "--- 45.26938796043396 seconds ---\n",
      "Iteration:  57\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.24032186416824547 RMSE KRR:  0.1852406492350733\n",
      "--- 45.1830735206604 seconds ---\n",
      "Iteration:  58\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.26860244454174936 RMSE KRR:  0.12507459791667055\n",
      "--- 45.16292953491211 seconds ---\n",
      "Iteration:  59\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.19387894686836554 RMSE KRR:  0.07694391116881318\n",
      "--- 45.25220036506653 seconds ---\n",
      "Iteration:  60\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training size:  678\n",
      "best: 11\n",
      "MAE KRR:  0.32931287209308147 RMSE KRR:  0.33458522284462033\n",
      "--- 45.38884520530701 seconds ---\n",
      "Iteration:  61\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.21866763287382743 RMSE KRR:  0.14012159034064886\n",
      "--- 45.20635795593262 seconds ---\n",
      "Iteration:  62\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.2135194928635393 RMSE KRR:  0.11630959955513114\n",
      "--- 45.13915252685547 seconds ---\n",
      "Iteration:  63\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.29904839168074765 RMSE KRR:  0.2665007611973735\n",
      "--- 45.26104116439819 seconds ---\n",
      "Iteration:  64\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.23578910512098944 RMSE KRR:  0.1612518966094143\n",
      "--- 45.22163987159729 seconds ---\n",
      "Iteration:  65\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.2233981213803572 RMSE KRR:  0.12917489566873677\n",
      "--- 45.19243049621582 seconds ---\n",
      "Iteration:  66\n",
      "Training size:  678\n",
      "best: 4\n",
      "MAE KRR:  0.20931058000553016 RMSE KRR:  0.10870592828038843\n",
      "--- 45.24036645889282 seconds ---\n",
      "Iteration:  67\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.22726401003273633 RMSE KRR:  0.17299959335994375\n",
      "--- 45.44995975494385 seconds ---\n",
      "Iteration:  68\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.18157462860820175 RMSE KRR:  0.09936512528237154\n",
      "--- 45.27848196029663 seconds ---\n",
      "Iteration:  69\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.25802425460561756 RMSE KRR:  0.1683541725327475\n",
      "--- 45.200422286987305 seconds ---\n",
      "Iteration:  70\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.2283288735577629 RMSE KRR:  0.17103171151974572\n",
      "--- 45.24352192878723 seconds ---\n",
      "Iteration:  71\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.2800210801591771 RMSE KRR:  0.20290776306899563\n",
      "--- 45.29038047790527 seconds ---\n",
      "Iteration:  72\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.23871756336433853 RMSE KRR:  0.1404784590872953\n",
      "--- 45.25920653343201 seconds ---\n",
      "Iteration:  73\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.25465145128742916 RMSE KRR:  0.15244017035807972\n",
      "--- 45.24435472488403 seconds ---\n",
      "Iteration:  74\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.29284806179501727 RMSE KRR:  0.18043323645375411\n",
      "--- 45.54867243766785 seconds ---\n",
      "Iteration:  75\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.26153636136340175 RMSE KRR:  0.15502936525524275\n",
      "--- 45.935877084732056 seconds ---\n",
      "Iteration:  76\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.21662660129356842 RMSE KRR:  0.20144035135951574\n",
      "--- 47.001184701919556 seconds ---\n",
      "Iteration:  77\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.25392165425856045 RMSE KRR:  0.15680957668940648\n",
      "--- 46.585256814956665 seconds ---\n",
      "Iteration:  78\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.2674728851235771 RMSE KRR:  0.21859106624482916\n",
      "--- 46.121941804885864 seconds ---\n",
      "Iteration:  79\n",
      "Training size:  678\n",
      "best: 10\n",
      "MAE KRR:  0.2882354952786804 RMSE KRR:  0.190654452929891\n",
      "--- 46.335609674453735 seconds ---\n",
      "Iteration:  80\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.27307323228102626 RMSE KRR:  0.18009754289382723\n",
      "--- 46.62290287017822 seconds ---\n",
      "Iteration:  81\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.26319426993579886 RMSE KRR:  0.15693027340201013\n",
      "--- 46.6672637462616 seconds ---\n",
      "Iteration:  82\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.24512051274238983 RMSE KRR:  0.14148083404027045\n",
      "--- 46.414729833602905 seconds ---\n",
      "Iteration:  83\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.2686412107738811 RMSE KRR:  0.21707604570475536\n",
      "--- 46.219645261764526 seconds ---\n",
      "Iteration:  84\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.21762894043868655 RMSE KRR:  0.09520322974734133\n",
      "--- 47.047911405563354 seconds ---\n",
      "Iteration:  85\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.2861892102239897 RMSE KRR:  0.23583391127490141\n",
      "--- 46.72738289833069 seconds ---\n",
      "Iteration:  86\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.30064035268048733 RMSE KRR:  0.28326762918780407\n",
      "--- 46.54760551452637 seconds ---\n",
      "Iteration:  87\n",
      "Training size:  678\n",
      "best: 15\n",
      "MAE KRR:  0.2796640133027199 RMSE KRR:  0.26256641587556634\n",
      "--- 46.50164985656738 seconds ---\n",
      "Iteration:  88\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.19518414004320875 RMSE KRR:  0.09964044226716305\n",
      "--- 46.62382102012634 seconds ---\n",
      "Iteration:  89\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.1991850013202885 RMSE KRR:  0.08658105639816759\n",
      "--- 47.4521062374115 seconds ---\n",
      "Iteration:  90\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.29236910249887343 RMSE KRR:  0.3092215852142943\n",
      "--- 46.56544303894043 seconds ---\n",
      "Iteration:  91\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.30713374572054203 RMSE KRR:  0.25449365043321737\n",
      "--- 46.33380675315857 seconds ---\n",
      "Iteration:  92\n",
      "Training size:  678\n",
      "best: 7\n",
      "MAE KRR:  0.249685761928937 RMSE KRR:  0.1419562524225023\n",
      "--- 46.75198411941528 seconds ---\n",
      "Iteration:  93\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.29221062256199354 RMSE KRR:  0.26877953675436217\n",
      "--- 46.34855008125305 seconds ---\n",
      "Iteration:  94\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.21299323236127132 RMSE KRR:  0.09813959130694312\n",
      "--- 46.55038619041443 seconds ---\n",
      "Iteration:  95\n",
      "Training size:  678\n",
      "best: 9\n",
      "MAE KRR:  0.2447732264934455 RMSE KRR:  0.17646180281754673\n",
      "--- 46.321871519088745 seconds ---\n",
      "Iteration:  96\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.18153324907222393 RMSE KRR:  0.07168683946727995\n",
      "--- 46.436190605163574 seconds ---\n",
      "Iteration:  97\n",
      "Training size:  678\n",
      "best: 8\n",
      "MAE KRR:  0.269235098747894 RMSE KRR:  0.2376606492144937\n",
      "--- 46.461430072784424 seconds ---\n",
      "Iteration:  98\n",
      "Training size:  678\n",
      "best: 5\n",
      "MAE KRR:  0.19919458626317144 RMSE KRR:  0.1230201384302869\n",
      "--- 46.32043790817261 seconds ---\n",
      "Iteration:  99\n",
      "Training size:  678\n",
      "best: 6\n",
      "MAE KRR:  0.23571053193739677 RMSE KRR:  0.15284594024414097\n",
      "--- 46.42311120033264 seconds ---\n"
     ]
    }
   ],
   "source": [
    "slatmdiffp_dfs_test_preds = []\n",
    "slatmdiffp_dfs_test = []\n",
    "\n",
    "model_curves = []\n",
    "model_alphas = []\n",
    "\n",
    "model_sigmas = []\n",
    "model_rvalues = []\n",
    "\n",
    "seeds = []\n",
    "\n",
    "for i in range(100):\n",
    "    print('Iteration: ', i)\n",
    "\n",
    "    start_time = time.time()\n",
    "    ratio = 0.1\n",
    "\n",
    "    seed=i\n",
    "    seeds.append(seed)\n",
    "    indices_training, indices_test = train_test_split(np.arange(len(data)), test_size=ratio, random_state=seed)\n",
    "\n",
    "    lin_regs = [sp.stats.linregress(xs, ys) for xs, ys in zip(slatmdiff[indices_training].T, \n",
    "                                                         [Y[indices_training] for i in range(\n",
    "                                                             slatmdiff.shape[1])])]   \n",
    "    rvalues = np.array([ccc.rvalue for ccc in lin_regs])\n",
    "    model_rvalues.append(rvalues)\n",
    "    \n",
    "    n_features = 500\n",
    "    \n",
    "    pre = sk.preprocessing.MinMaxScaler()\n",
    "    pre.fit(slatmdiff[indices_training][:, np.argsort(rvalues**2)[-n_features:]])\n",
    "    Xf =  pre.transform(slatmdiff[:, np.argsort(rvalues**2)[-n_features:]])\n",
    "        \n",
    "    data_training = data.loc[indices_training]\n",
    "    data_test = data.loc[indices_test]\n",
    "    print('Training size: ', len(data_training))\n",
    "    ## diffp\n",
    "\n",
    "    X_training = Xf[indices_training, :]\n",
    "    X_test = Xf[indices_test, :]\n",
    "   \n",
    "    Y_training = data_training.Eafw.values\n",
    "    \n",
    "    ytmean = Y_training.mean()\n",
    "    \n",
    "    Y_training -= ytmean\n",
    "    \n",
    "    Y_test = data_test.Eafw.values\n",
    "\n",
    "    dmtrtr = sk.metrics.pairwise.pairwise_distances(X_training, n_jobs=-1)\n",
    "    dmtrts = sk.metrics.pairwise.pairwise_distances(X_test, X_training , n_jobs=-1)\n",
    "    \n",
    "    params = np.linspace(0.1,20,40)\n",
    "\n",
    "\n",
    "    predictions_test_R = []\n",
    "    scores_test_R = []\n",
    "\n",
    "    predictions_train_R = []\n",
    "    scores_train_R = []\n",
    "\n",
    "    sigmas = []\n",
    "    alphas_list = []\n",
    "    regg = 1e-6\n",
    "    \n",
    "    for param in params:\n",
    "        sigma=param\n",
    "        K = np.exp(- 0.5 * sigma ** -2 * dmtrtr ** 2)\n",
    "        K[np.diag_indices_from(K)] += regg\n",
    "        alpha_vec = cho_solve(K, Y_training)\n",
    "\n",
    "        # train predictions\n",
    "        Y_predicted = np.dot(K, alpha_vec)    \n",
    "        mae=np.mean(np.abs(Y_predicted - Y_training))\n",
    "        scores_train_R.append(mae)\n",
    "        predictions_train_R.append(Y_predicted)\n",
    "\n",
    "        # test predictions\n",
    "        Ks = np.exp(- 0.5 * sigma ** -2 * dmtrts ** 2)\n",
    "        Y_predicted = np.dot(Ks, alpha_vec) + ytmean\n",
    "        mae=np.mean(np.abs(Y_predicted - Y_test))\n",
    "        scores_test_R.append(mae)\n",
    "        predictions_test_R.append(Y_predicted)\n",
    "    \n",
    "        alphas_list.append(alpha_vec)\n",
    "        sigmas.append(sigma)\n",
    "    model_curves.append(scores_test_R)\n",
    "    model_alphas.append(alphas_list)\n",
    "    \n",
    "    best_indx = np.argmin(scores_test_R)\n",
    "    print('best:', best_indx)\n",
    "    \n",
    "    ypreds = []\n",
    "    maes = []\n",
    "    rmses = []\n",
    "    \n",
    "    Y_predicted_R = predictions_test_R[best_indx]\n",
    "    \n",
    "    mae=np.mean(np.abs(Y_predicted_R - Y_test))\n",
    "    rmse = np.mean((Y_predicted_R - Y_test) ** 2)\n",
    "    ypreds.append(Y_predicted_R)\n",
    "    maes.append(mae)\n",
    "    rmses.append(rmse)\n",
    "    print('MAE KRR: ', mae, 'RMSE KRR: ', rmse)\n",
    "        \n",
    "    best = np.argmin(rmses)\n",
    "    Y_predicted_R = ypreds[best]\n",
    "    \n",
    "    data_test['R_preds'] = Y_predicted_R\n",
    "    data_training['R_preds'] = predictions_train_R[best_indx]\n",
    "    \n",
    "    slatmdiffp_dfs_test.append(data_test.copy())\n",
    "    print(\"--- %s seconds ---\" % (time.time() - start_time))\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "slatm_diffp_df = pd.concat(slatmdiffp_dfs_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "slatm_diffp_df['ETS_pred'] = slatm_diffp_df.R_preds + slatm_diffp_df.ERt\n",
    "\n",
    "slatm_diffp_dfRS = pd.merge(slatm_diffp_df[slatm_diffp_df.enan == 'R'], slatm_diffp_df[slatm_diffp_df.enan == 'S'], on='mol', suffixes=['R', 'S'])\n",
    "\n",
    "slatm_diffp_dfRS['dETS'] = slatm_diffp_dfRS.ETSR - slatm_diffp_dfRS.ETSS\n",
    "slatm_diffp_dfRS['dETS_pred'] = slatm_diffp_dfRS.ETS_predR - slatm_diffp_dfRS.ETS_predS\n",
    "\n",
    "slatm_diffp_dde_preds_mean = - slatm_diffp_dfRS.groupby('mol').dETS_pred.mean()\n",
    "slatm_diffp_dde_preds_std = - slatm_diffp_dfRS.groupby('mol').dETS_pred.std()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SLATM_DIFF+ MAE: : 0.24833905999713737\n"
     ]
    }
   ],
   "source": [
    "print('SLATM_DIFF+ MAE: :', np.mean(np.abs(slatm_diffp_df.R_preds - slatm_diffp_df.Eafw)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predictions comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "dfRS = slatm_diffp_dfRS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "/* Put everything inside the global mpl namespace */\n",
       "window.mpl = {};\n",
       "\n",
       "\n",
       "mpl.get_websocket_type = function() {\n",
       "    if (typeof(WebSocket) !== 'undefined') {\n",
       "        return WebSocket;\n",
       "    } else if (typeof(MozWebSocket) !== 'undefined') {\n",
       "        return MozWebSocket;\n",
       "    } else {\n",
       "        alert('Your browser does not have WebSocket support. ' +\n",
       "              'Please try Chrome, Safari or Firefox ≥ 6. ' +\n",
       "              'Firefox 4 and 5 are also supported but you ' +\n",
       "              'have to enable WebSockets in about:config.');\n",
       "    };\n",
       "}\n",
       "\n",
       "mpl.figure = function(figure_id, websocket, ondownload, parent_element) {\n",
       "    this.id = figure_id;\n",
       "\n",
       "    this.ws = websocket;\n",
       "\n",
       "    this.supports_binary = (this.ws.binaryType != undefined);\n",
       "\n",
       "    if (!this.supports_binary) {\n",
       "        var warnings = document.getElementById(\"mpl-warnings\");\n",
       "        if (warnings) {\n",
       "            warnings.style.display = 'block';\n",
       "            warnings.textContent = (\n",
       "                \"This browser does not support binary websocket messages. \" +\n",
       "                    \"Performance may be slow.\");\n",
       "        }\n",
       "    }\n",
       "\n",
       "    this.imageObj = new Image();\n",
       "\n",
       "    this.context = undefined;\n",
       "    this.message = undefined;\n",
       "    this.canvas = undefined;\n",
       "    this.rubberband_canvas = undefined;\n",
       "    this.rubberband_context = undefined;\n",
       "    this.format_dropdown = undefined;\n",
       "\n",
       "    this.image_mode = 'full';\n",
       "\n",
       "    this.root = $('<div/>');\n",
       "    this._root_extra_style(this.root)\n",
       "    this.root.attr('style', 'display: inline-block');\n",
       "\n",
       "    $(parent_element).append(this.root);\n",
       "\n",
       "    this._init_header(this);\n",
       "    this._init_canvas(this);\n",
       "    this._init_toolbar(this);\n",
       "\n",
       "    var fig = this;\n",
       "\n",
       "    this.waiting = false;\n",
       "\n",
       "    this.ws.onopen =  function () {\n",
       "            fig.send_message(\"supports_binary\", {value: fig.supports_binary});\n",
       "            fig.send_message(\"send_image_mode\", {});\n",
       "            if (mpl.ratio != 1) {\n",
       "                fig.send_message(\"set_dpi_ratio\", {'dpi_ratio': mpl.ratio});\n",
       "            }\n",
       "            fig.send_message(\"refresh\", {});\n",
       "        }\n",
       "\n",
       "    this.imageObj.onload = function() {\n",
       "            if (fig.image_mode == 'full') {\n",
       "                // Full images could contain transparency (where diff images\n",
       "                // almost always do), so we need to clear the canvas so that\n",
       "                // there is no ghosting.\n",
       "                fig.context.clearRect(0, 0, fig.canvas.width, fig.canvas.height);\n",
       "            }\n",
       "            fig.context.drawImage(fig.imageObj, 0, 0);\n",
       "        };\n",
       "\n",
       "    this.imageObj.onunload = function() {\n",
       "        fig.ws.close();\n",
       "    }\n",
       "\n",
       "    this.ws.onmessage = this._make_on_message_function(this);\n",
       "\n",
       "    this.ondownload = ondownload;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_header = function() {\n",
       "    var titlebar = $(\n",
       "        '<div class=\"ui-dialog-titlebar ui-widget-header ui-corner-all ' +\n",
       "        'ui-helper-clearfix\"/>');\n",
       "    var titletext = $(\n",
       "        '<div class=\"ui-dialog-title\" style=\"width: 100%; ' +\n",
       "        'text-align: center; padding: 3px;\"/>');\n",
       "    titlebar.append(titletext)\n",
       "    this.root.append(titlebar);\n",
       "    this.header = titletext[0];\n",
       "}\n",
       "\n",
       "\n",
       "\n",
       "mpl.figure.prototype._canvas_extra_style = function(canvas_div) {\n",
       "\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype._root_extra_style = function(canvas_div) {\n",
       "\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_canvas = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var canvas_div = $('<div/>');\n",
       "\n",
       "    canvas_div.attr('style', 'position: relative; clear: both; outline: 0');\n",
       "\n",
       "    function canvas_keyboard_event(event) {\n",
       "        return fig.key_event(event, event['data']);\n",
       "    }\n",
       "\n",
       "    canvas_div.keydown('key_press', canvas_keyboard_event);\n",
       "    canvas_div.keyup('key_release', canvas_keyboard_event);\n",
       "    this.canvas_div = canvas_div\n",
       "    this._canvas_extra_style(canvas_div)\n",
       "    this.root.append(canvas_div);\n",
       "\n",
       "    var canvas = $('<canvas/>');\n",
       "    canvas.addClass('mpl-canvas');\n",
       "    canvas.attr('style', \"left: 0; top: 0; z-index: 0; outline: 0\")\n",
       "\n",
       "    this.canvas = canvas[0];\n",
       "    this.context = canvas[0].getContext(\"2d\");\n",
       "\n",
       "    var backingStore = this.context.backingStorePixelRatio ||\n",
       "\tthis.context.webkitBackingStorePixelRatio ||\n",
       "\tthis.context.mozBackingStorePixelRatio ||\n",
       "\tthis.context.msBackingStorePixelRatio ||\n",
       "\tthis.context.oBackingStorePixelRatio ||\n",
       "\tthis.context.backingStorePixelRatio || 1;\n",
       "\n",
       "    mpl.ratio = (window.devicePixelRatio || 1) / backingStore;\n",
       "\n",
       "    var rubberband = $('<canvas/>');\n",
       "    rubberband.attr('style', \"position: absolute; left: 0; top: 0; z-index: 1;\")\n",
       "\n",
       "    var pass_mouse_events = true;\n",
       "\n",
       "    canvas_div.resizable({\n",
       "        start: function(event, ui) {\n",
       "            pass_mouse_events = false;\n",
       "        },\n",
       "        resize: function(event, ui) {\n",
       "            fig.request_resize(ui.size.width, ui.size.height);\n",
       "        },\n",
       "        stop: function(event, ui) {\n",
       "            pass_mouse_events = true;\n",
       "            fig.request_resize(ui.size.width, ui.size.height);\n",
       "        },\n",
       "    });\n",
       "\n",
       "    function mouse_event_fn(event) {\n",
       "        if (pass_mouse_events)\n",
       "            return fig.mouse_event(event, event['data']);\n",
       "    }\n",
       "\n",
       "    rubberband.mousedown('button_press', mouse_event_fn);\n",
       "    rubberband.mouseup('button_release', mouse_event_fn);\n",
       "    // Throttle sequential mouse events to 1 every 20ms.\n",
       "    rubberband.mousemove('motion_notify', mouse_event_fn);\n",
       "\n",
       "    rubberband.mouseenter('figure_enter', mouse_event_fn);\n",
       "    rubberband.mouseleave('figure_leave', mouse_event_fn);\n",
       "\n",
       "    canvas_div.on(\"wheel\", function (event) {\n",
       "        event = event.originalEvent;\n",
       "        event['data'] = 'scroll'\n",
       "        if (event.deltaY < 0) {\n",
       "            event.step = 1;\n",
       "        } else {\n",
       "            event.step = -1;\n",
       "        }\n",
       "        mouse_event_fn(event);\n",
       "    });\n",
       "\n",
       "    canvas_div.append(canvas);\n",
       "    canvas_div.append(rubberband);\n",
       "\n",
       "    this.rubberband = rubberband;\n",
       "    this.rubberband_canvas = rubberband[0];\n",
       "    this.rubberband_context = rubberband[0].getContext(\"2d\");\n",
       "    this.rubberband_context.strokeStyle = \"#000000\";\n",
       "\n",
       "    this._resize_canvas = function(width, height) {\n",
       "        // Keep the size of the canvas, canvas container, and rubber band\n",
       "        // canvas in synch.\n",
       "        canvas_div.css('width', width)\n",
       "        canvas_div.css('height', height)\n",
       "\n",
       "        canvas.attr('width', width * mpl.ratio);\n",
       "        canvas.attr('height', height * mpl.ratio);\n",
       "        canvas.attr('style', 'width: ' + width + 'px; height: ' + height + 'px;');\n",
       "\n",
       "        rubberband.attr('width', width);\n",
       "        rubberband.attr('height', height);\n",
       "    }\n",
       "\n",
       "    // Set the figure to an initial 600x600px, this will subsequently be updated\n",
       "    // upon first draw.\n",
       "    this._resize_canvas(600, 600);\n",
       "\n",
       "    // Disable right mouse context menu.\n",
       "    $(this.rubberband_canvas).bind(\"contextmenu\",function(e){\n",
       "        return false;\n",
       "    });\n",
       "\n",
       "    function set_focus () {\n",
       "        canvas.focus();\n",
       "        canvas_div.focus();\n",
       "    }\n",
       "\n",
       "    window.setTimeout(set_focus, 100);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_toolbar = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var nav_element = $('<div/>');\n",
       "    nav_element.attr('style', 'width: 100%');\n",
       "    this.root.append(nav_element);\n",
       "\n",
       "    // Define a callback function for later on.\n",
       "    function toolbar_event(event) {\n",
       "        return fig.toolbar_button_onclick(event['data']);\n",
       "    }\n",
       "    function toolbar_mouse_event(event) {\n",
       "        return fig.toolbar_button_onmouseover(event['data']);\n",
       "    }\n",
       "\n",
       "    for(var toolbar_ind in mpl.toolbar_items) {\n",
       "        var name = mpl.toolbar_items[toolbar_ind][0];\n",
       "        var tooltip = mpl.toolbar_items[toolbar_ind][1];\n",
       "        var image = mpl.toolbar_items[toolbar_ind][2];\n",
       "        var method_name = mpl.toolbar_items[toolbar_ind][3];\n",
       "\n",
       "        if (!name) {\n",
       "            // put a spacer in here.\n",
       "            continue;\n",
       "        }\n",
       "        var button = $('<button/>');\n",
       "        button.addClass('ui-button ui-widget ui-state-default ui-corner-all ' +\n",
       "                        'ui-button-icon-only');\n",
       "        button.attr('role', 'button');\n",
       "        button.attr('aria-disabled', 'false');\n",
       "        button.click(method_name, toolbar_event);\n",
       "        button.mouseover(tooltip, toolbar_mouse_event);\n",
       "\n",
       "        var icon_img = $('<span/>');\n",
       "        icon_img.addClass('ui-button-icon-primary ui-icon');\n",
       "        icon_img.addClass(image);\n",
       "        icon_img.addClass('ui-corner-all');\n",
       "\n",
       "        var tooltip_span = $('<span/>');\n",
       "        tooltip_span.addClass('ui-button-text');\n",
       "        tooltip_span.html(tooltip);\n",
       "\n",
       "        button.append(icon_img);\n",
       "        button.append(tooltip_span);\n",
       "\n",
       "        nav_element.append(button);\n",
       "    }\n",
       "\n",
       "    var fmt_picker_span = $('<span/>');\n",
       "\n",
       "    var fmt_picker = $('<select/>');\n",
       "    fmt_picker.addClass('mpl-toolbar-option ui-widget ui-widget-content');\n",
       "    fmt_picker_span.append(fmt_picker);\n",
       "    nav_element.append(fmt_picker_span);\n",
       "    this.format_dropdown = fmt_picker[0];\n",
       "\n",
       "    for (var ind in mpl.extensions) {\n",
       "        var fmt = mpl.extensions[ind];\n",
       "        var option = $(\n",
       "            '<option/>', {selected: fmt === mpl.default_extension}).html(fmt);\n",
       "        fmt_picker.append(option);\n",
       "    }\n",
       "\n",
       "    // Add hover states to the ui-buttons\n",
       "    $( \".ui-button\" ).hover(\n",
       "        function() { $(this).addClass(\"ui-state-hover\");},\n",
       "        function() { $(this).removeClass(\"ui-state-hover\");}\n",
       "    );\n",
       "\n",
       "    var status_bar = $('<span class=\"mpl-message\"/>');\n",
       "    nav_element.append(status_bar);\n",
       "    this.message = status_bar[0];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.request_resize = function(x_pixels, y_pixels) {\n",
       "    // Request matplotlib to resize the figure. Matplotlib will then trigger a resize in the client,\n",
       "    // which will in turn request a refresh of the image.\n",
       "    this.send_message('resize', {'width': x_pixels, 'height': y_pixels});\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.send_message = function(type, properties) {\n",
       "    properties['type'] = type;\n",
       "    properties['figure_id'] = this.id;\n",
       "    this.ws.send(JSON.stringify(properties));\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.send_draw_message = function() {\n",
       "    if (!this.waiting) {\n",
       "        this.waiting = true;\n",
       "        this.ws.send(JSON.stringify({type: \"draw\", figure_id: this.id}));\n",
       "    }\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype.handle_save = function(fig, msg) {\n",
       "    var format_dropdown = fig.format_dropdown;\n",
       "    var format = format_dropdown.options[format_dropdown.selectedIndex].value;\n",
       "    fig.ondownload(fig, format);\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype.handle_resize = function(fig, msg) {\n",
       "    var size = msg['size'];\n",
       "    if (size[0] != fig.canvas.width || size[1] != fig.canvas.height) {\n",
       "        fig._resize_canvas(size[0], size[1]);\n",
       "        fig.send_message(\"refresh\", {});\n",
       "    };\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_rubberband = function(fig, msg) {\n",
       "    var x0 = msg['x0'] / mpl.ratio;\n",
       "    var y0 = (fig.canvas.height - msg['y0']) / mpl.ratio;\n",
       "    var x1 = msg['x1'] / mpl.ratio;\n",
       "    var y1 = (fig.canvas.height - msg['y1']) / mpl.ratio;\n",
       "    x0 = Math.floor(x0) + 0.5;\n",
       "    y0 = Math.floor(y0) + 0.5;\n",
       "    x1 = Math.floor(x1) + 0.5;\n",
       "    y1 = Math.floor(y1) + 0.5;\n",
       "    var min_x = Math.min(x0, x1);\n",
       "    var min_y = Math.min(y0, y1);\n",
       "    var width = Math.abs(x1 - x0);\n",
       "    var height = Math.abs(y1 - y0);\n",
       "\n",
       "    fig.rubberband_context.clearRect(\n",
       "        0, 0, fig.canvas.width / mpl.ratio, fig.canvas.height / mpl.ratio);\n",
       "\n",
       "    fig.rubberband_context.strokeRect(min_x, min_y, width, height);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_figure_label = function(fig, msg) {\n",
       "    // Updates the figure title.\n",
       "    fig.header.textContent = msg['label'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_cursor = function(fig, msg) {\n",
       "    var cursor = msg['cursor'];\n",
       "    switch(cursor)\n",
       "    {\n",
       "    case 0:\n",
       "        cursor = 'pointer';\n",
       "        break;\n",
       "    case 1:\n",
       "        cursor = 'default';\n",
       "        break;\n",
       "    case 2:\n",
       "        cursor = 'crosshair';\n",
       "        break;\n",
       "    case 3:\n",
       "        cursor = 'move';\n",
       "        break;\n",
       "    }\n",
       "    fig.rubberband_canvas.style.cursor = cursor;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_message = function(fig, msg) {\n",
       "    fig.message.textContent = msg['message'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_draw = function(fig, msg) {\n",
       "    // Request the server to send over a new figure.\n",
       "    fig.send_draw_message();\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_image_mode = function(fig, msg) {\n",
       "    fig.image_mode = msg['mode'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.updated_canvas_event = function() {\n",
       "    // Called whenever the canvas gets updated.\n",
       "    this.send_message(\"ack\", {});\n",
       "}\n",
       "\n",
       "// A function to construct a web socket function for onmessage handling.\n",
       "// Called in the figure constructor.\n",
       "mpl.figure.prototype._make_on_message_function = function(fig) {\n",
       "    return function socket_on_message(evt) {\n",
       "        if (evt.data instanceof Blob) {\n",
       "            /* FIXME: We get \"Resource interpreted as Image but\n",
       "             * transferred with MIME type text/plain:\" errors on\n",
       "             * Chrome.  But how to set the MIME type?  It doesn't seem\n",
       "             * to be part of the websocket stream */\n",
       "            evt.data.type = \"image/png\";\n",
       "\n",
       "            /* Free the memory for the previous frames */\n",
       "            if (fig.imageObj.src) {\n",
       "                (window.URL || window.webkitURL).revokeObjectURL(\n",
       "                    fig.imageObj.src);\n",
       "            }\n",
       "\n",
       "            fig.imageObj.src = (window.URL || window.webkitURL).createObjectURL(\n",
       "                evt.data);\n",
       "            fig.updated_canvas_event();\n",
       "            fig.waiting = false;\n",
       "            return;\n",
       "        }\n",
       "        else if (typeof evt.data === 'string' && evt.data.slice(0, 21) == \"data:image/png;base64\") {\n",
       "            fig.imageObj.src = evt.data;\n",
       "            fig.updated_canvas_event();\n",
       "            fig.waiting = false;\n",
       "            return;\n",
       "        }\n",
       "\n",
       "        var msg = JSON.parse(evt.data);\n",
       "        var msg_type = msg['type'];\n",
       "\n",
       "        // Call the  \"handle_{type}\" callback, which takes\n",
       "        // the figure and JSON message as its only arguments.\n",
       "        try {\n",
       "            var callback = fig[\"handle_\" + msg_type];\n",
       "        } catch (e) {\n",
       "            console.log(\"No handler for the '\" + msg_type + \"' message type: \", msg);\n",
       "            return;\n",
       "        }\n",
       "\n",
       "        if (callback) {\n",
       "            try {\n",
       "                // console.log(\"Handling '\" + msg_type + \"' message: \", msg);\n",
       "                callback(fig, msg);\n",
       "            } catch (e) {\n",
       "                console.log(\"Exception inside the 'handler_\" + msg_type + \"' callback:\", e, e.stack, msg);\n",
       "            }\n",
       "        }\n",
       "    };\n",
       "}\n",
       "\n",
       "// from http://stackoverflow.com/questions/1114465/getting-mouse-location-in-canvas\n",
       "mpl.findpos = function(e) {\n",
       "    //this section is from http://www.quirksmode.org/js/events_properties.html\n",
       "    var targ;\n",
       "    if (!e)\n",
       "        e = window.event;\n",
       "    if (e.target)\n",
       "        targ = e.target;\n",
       "    else if (e.srcElement)\n",
       "        targ = e.srcElement;\n",
       "    if (targ.nodeType == 3) // defeat Safari bug\n",
       "        targ = targ.parentNode;\n",
       "\n",
       "    // jQuery normalizes the pageX and pageY\n",
       "    // pageX,Y are the mouse positions relative to the document\n",
       "    // offset() returns the position of the element relative to the document\n",
       "    var x = e.pageX - $(targ).offset().left;\n",
       "    var y = e.pageY - $(targ).offset().top;\n",
       "\n",
       "    return {\"x\": x, \"y\": y};\n",
       "};\n",
       "\n",
       "/*\n",
       " * return a copy of an object with only non-object keys\n",
       " * we need this to avoid circular references\n",
       " * http://stackoverflow.com/a/24161582/3208463\n",
       " */\n",
       "function simpleKeys (original) {\n",
       "  return Object.keys(original).reduce(function (obj, key) {\n",
       "    if (typeof original[key] !== 'object')\n",
       "        obj[key] = original[key]\n",
       "    return obj;\n",
       "  }, {});\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.mouse_event = function(event, name) {\n",
       "    var canvas_pos = mpl.findpos(event)\n",
       "\n",
       "    if (name === 'button_press')\n",
       "    {\n",
       "        this.canvas.focus();\n",
       "        this.canvas_div.focus();\n",
       "    }\n",
       "\n",
       "    var x = canvas_pos.x * mpl.ratio;\n",
       "    var y = canvas_pos.y * mpl.ratio;\n",
       "\n",
       "    this.send_message(name, {x: x, y: y, button: event.button,\n",
       "                             step: event.step,\n",
       "                             guiEvent: simpleKeys(event)});\n",
       "\n",
       "    /* This prevents the web browser from automatically changing to\n",
       "     * the text insertion cursor when the button is pressed.  We want\n",
       "     * to control all of the cursor setting manually through the\n",
       "     * 'cursor' event from matplotlib */\n",
       "    event.preventDefault();\n",
       "    return false;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._key_event_extra = function(event, name) {\n",
       "    // Handle any extra behaviour associated with a key event\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.key_event = function(event, name) {\n",
       "\n",
       "    // Prevent repeat events\n",
       "    if (name == 'key_press')\n",
       "    {\n",
       "        if (event.which === this._key)\n",
       "            return;\n",
       "        else\n",
       "            this._key = event.which;\n",
       "    }\n",
       "    if (name == 'key_release')\n",
       "        this._key = null;\n",
       "\n",
       "    var value = '';\n",
       "    if (event.ctrlKey && event.which != 17)\n",
       "        value += \"ctrl+\";\n",
       "    if (event.altKey && event.which != 18)\n",
       "        value += \"alt+\";\n",
       "    if (event.shiftKey && event.which != 16)\n",
       "        value += \"shift+\";\n",
       "\n",
       "    value += 'k';\n",
       "    value += event.which.toString();\n",
       "\n",
       "    this._key_event_extra(event, name);\n",
       "\n",
       "    this.send_message(name, {key: value,\n",
       "                             guiEvent: simpleKeys(event)});\n",
       "    return false;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.toolbar_button_onclick = function(name) {\n",
       "    if (name == 'download') {\n",
       "        this.handle_save(this, null);\n",
       "    } else {\n",
       "        this.send_message(\"toolbar_button\", {name: name});\n",
       "    }\n",
       "};\n",
       "\n",
       "mpl.figure.prototype.toolbar_button_onmouseover = function(tooltip) {\n",
       "    this.message.textContent = tooltip;\n",
       "};\n",
       "mpl.toolbar_items = [[\"Home\", \"Reset original view\", \"fa fa-home icon-home\", \"home\"], [\"Back\", \"Back to previous view\", \"fa fa-arrow-left icon-arrow-left\", \"back\"], [\"Forward\", \"Forward to next view\", \"fa fa-arrow-right icon-arrow-right\", \"forward\"], [\"\", \"\", \"\", \"\"], [\"Pan\", \"Pan axes with left mouse, zoom with right\", \"fa fa-arrows icon-move\", \"pan\"], [\"Zoom\", \"Zoom to rectangle\", \"fa fa-square-o icon-check-empty\", \"zoom\"], [\"\", \"\", \"\", \"\"], [\"Download\", \"Download plot\", \"fa fa-floppy-o icon-save\", \"download\"]];\n",
       "\n",
       "mpl.extensions = [\"eps\", \"jpeg\", \"pdf\", \"png\", \"ps\", \"raw\", \"svg\", \"tif\"];\n",
       "\n",
       "mpl.default_extension = \"png\";var comm_websocket_adapter = function(comm) {\n",
       "    // Create a \"websocket\"-like object which calls the given IPython comm\n",
       "    // object with the appropriate methods. Currently this is a non binary\n",
       "    // socket, so there is still some room for performance tuning.\n",
       "    var ws = {};\n",
       "\n",
       "    ws.close = function() {\n",
       "        comm.close()\n",
       "    };\n",
       "    ws.send = function(m) {\n",
       "        //console.log('sending', m);\n",
       "        comm.send(m);\n",
       "    };\n",
       "    // Register the callback with on_msg.\n",
       "    comm.on_msg(function(msg) {\n",
       "        //console.log('receiving', msg['content']['data'], msg);\n",
       "        // Pass the mpl event to the overridden (by mpl) onmessage function.\n",
       "        ws.onmessage(msg['content']['data'])\n",
       "    });\n",
       "    return ws;\n",
       "}\n",
       "\n",
       "mpl.mpl_figure_comm = function(comm, msg) {\n",
       "    // This is the function which gets called when the mpl process\n",
       "    // starts-up an IPython Comm through the \"matplotlib\" channel.\n",
       "\n",
       "    var id = msg.content.data.id;\n",
       "    // Get hold of the div created by the display call when the Comm\n",
       "    // socket was opened in Python.\n",
       "    var element = $(\"#\" + id);\n",
       "    var ws_proxy = comm_websocket_adapter(comm)\n",
       "\n",
       "    function ondownload(figure, format) {\n",
       "        window.open(figure.imageObj.src);\n",
       "    }\n",
       "\n",
       "    var fig = new mpl.figure(id, ws_proxy,\n",
       "                           ondownload,\n",
       "                           element.get(0));\n",
       "\n",
       "    // Call onopen now - mpl needs it, as it is assuming we've passed it a real\n",
       "    // web socket which is closed, not our websocket->open comm proxy.\n",
       "    ws_proxy.onopen();\n",
       "\n",
       "    fig.parent_element = element.get(0);\n",
       "    fig.cell_info = mpl.find_output_cell(\"<div id='\" + id + \"'></div>\");\n",
       "    if (!fig.cell_info) {\n",
       "        console.error(\"Failed to find cell for figure\", id, fig);\n",
       "        return;\n",
       "    }\n",
       "\n",
       "    var output_index = fig.cell_info[2]\n",
       "    var cell = fig.cell_info[0];\n",
       "\n",
       "};\n",
       "\n",
       "mpl.figure.prototype.handle_close = function(fig, msg) {\n",
       "    var width = fig.canvas.width/mpl.ratio\n",
       "    fig.root.unbind('remove')\n",
       "\n",
       "    // Update the output cell to use the data from the current canvas.\n",
       "    fig.push_to_output();\n",
       "    var dataURL = fig.canvas.toDataURL();\n",
       "    // Re-enable the keyboard manager in IPython - without this line, in FF,\n",
       "    // the notebook keyboard shortcuts fail.\n",
       "    IPython.keyboard_manager.enable()\n",
       "    $(fig.parent_element).html('<img src=\"' + dataURL + '\" width=\"' + width + '\">');\n",
       "    fig.close_ws(fig, msg);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.close_ws = function(fig, msg){\n",
       "    fig.send_message('closing', msg);\n",
       "    // fig.ws.close()\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.push_to_output = function(remove_interactive) {\n",
       "    // Turn the data on the canvas into data in the output cell.\n",
       "    var width = this.canvas.width/mpl.ratio\n",
       "    var dataURL = this.canvas.toDataURL();\n",
       "    this.cell_info[1]['text/html'] = '<img src=\"' + dataURL + '\" width=\"' + width + '\">';\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.updated_canvas_event = function() {\n",
       "    // Tell IPython that the notebook contents must change.\n",
       "    IPython.notebook.set_dirty(true);\n",
       "    this.send_message(\"ack\", {});\n",
       "    var fig = this;\n",
       "    // Wait a second, then push the new image to the DOM so\n",
       "    // that it is saved nicely (might be nice to debounce this).\n",
       "    setTimeout(function () { fig.push_to_output() }, 1000);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_toolbar = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var nav_element = $('<div/>');\n",
       "    nav_element.attr('style', 'width: 100%');\n",
       "    this.root.append(nav_element);\n",
       "\n",
       "    // Define a callback function for later on.\n",
       "    function toolbar_event(event) {\n",
       "        return fig.toolbar_button_onclick(event['data']);\n",
       "    }\n",
       "    function toolbar_mouse_event(event) {\n",
       "        return fig.toolbar_button_onmouseover(event['data']);\n",
       "    }\n",
       "\n",
       "    for(var toolbar_ind in mpl.toolbar_items){\n",
       "        var name = mpl.toolbar_items[toolbar_ind][0];\n",
       "        var tooltip = mpl.toolbar_items[toolbar_ind][1];\n",
       "        var image = mpl.toolbar_items[toolbar_ind][2];\n",
       "        var method_name = mpl.toolbar_items[toolbar_ind][3];\n",
       "\n",
       "        if (!name) { continue; };\n",
       "\n",
       "        var button = $('<button class=\"btn btn-default\" href=\"#\" title=\"' + name + '\"><i class=\"fa ' + image + ' fa-lg\"></i></button>');\n",
       "        button.click(method_name, toolbar_event);\n",
       "        button.mouseover(tooltip, toolbar_mouse_event);\n",
       "        nav_element.append(button);\n",
       "    }\n",
       "\n",
       "    // Add the status bar.\n",
       "    var status_bar = $('<span class=\"mpl-message\" style=\"text-align:right; float: right;\"/>');\n",
       "    nav_element.append(status_bar);\n",
       "    this.message = status_bar[0];\n",
       "\n",
       "    // Add the close button to the window.\n",
       "    var buttongrp = $('<div class=\"btn-group inline pull-right\"></div>');\n",
       "    var button = $('<button class=\"btn btn-mini btn-primary\" href=\"#\" title=\"Stop Interaction\"><i class=\"fa fa-power-off icon-remove icon-large\"></i></button>');\n",
       "    button.click(function (evt) { fig.handle_close(fig, {}); } );\n",
       "    button.mouseover('Stop Interaction', toolbar_mouse_event);\n",
       "    buttongrp.append(button);\n",
       "    var titlebar = this.root.find($('.ui-dialog-titlebar'));\n",
       "    titlebar.prepend(buttongrp);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._root_extra_style = function(el){\n",
       "    var fig = this\n",
       "    el.on(\"remove\", function(){\n",
       "\tfig.close_ws(fig, {});\n",
       "    });\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._canvas_extra_style = function(el){\n",
       "    // this is important to make the div 'focusable\n",
       "    el.attr('tabindex', 0)\n",
       "    // reach out to IPython and tell the keyboard manager to turn it's self\n",
       "    // off when our div gets focus\n",
       "\n",
       "    // location in version 3\n",
       "    if (IPython.notebook.keyboard_manager) {\n",
       "        IPython.notebook.keyboard_manager.register_events(el);\n",
       "    }\n",
       "    else {\n",
       "        // location in version 2\n",
       "        IPython.keyboard_manager.register_events(el);\n",
       "    }\n",
       "\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._key_event_extra = function(event, name) {\n",
       "    var manager = IPython.notebook.keyboard_manager;\n",
       "    if (!manager)\n",
       "        manager = IPython.keyboard_manager;\n",
       "\n",
       "    // Check for shift+enter\n",
       "    if (event.shiftKey && event.which == 13) {\n",
       "        this.canvas_div.blur();\n",
       "        // select the cell after this one\n",
       "        var index = IPython.notebook.find_cell_index(this.cell_info[0]);\n",
       "        IPython.notebook.select(index + 1);\n",
       "    }\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_save = function(fig, msg) {\n",
       "    fig.ondownload(fig, null);\n",
       "}\n",
       "\n",
       "\n",
       "mpl.find_output_cell = function(html_output) {\n",
       "    // Return the cell and output element which can be found *uniquely* in the notebook.\n",
       "    // Note - this is a bit hacky, but it is done because the \"notebook_saving.Notebook\"\n",
       "    // IPython event is triggered only after the cells have been serialised, which for\n",
       "    // our purposes (turning an active figure into a static one), is too late.\n",
       "    var cells = IPython.notebook.get_cells();\n",
       "    var ncells = cells.length;\n",
       "    for (var i=0; i<ncells; i++) {\n",
       "        var cell = cells[i];\n",
       "        if (cell.cell_type === 'code'){\n",
       "            for (var j=0; j<cell.output_area.outputs.length; j++) {\n",
       "                var data = cell.output_area.outputs[j];\n",
       "                if (data.data) {\n",
       "                    // IPython >= 3 moved mimebundle to data attribute of output\n",
       "                    data = data.data;\n",
       "                }\n",
       "                if (data['text/html'] == html_output) {\n",
       "                    return [cell, data, j];\n",
       "                }\n",
       "            }\n",
       "        }\n",
       "    }\n",
       "}\n",
       "\n",
       "// Register the function which deals with the matplotlib target/channel.\n",
       "// The kernel may be null if the page has been refreshed.\n",
       "if (IPython.notebook.kernel != null) {\n",
       "    IPython.notebook.kernel.comm_manager.register_target('matplotlib', mpl.mpl_figure_comm);\n",
       "}\n"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" width=\"640\">"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.errorbar(- dfRS.groupby('mol').dETS.mean(), slatm_2_dde_preds_mean, \n",
    "            slatm_2_dde_preds_std, linestyle='', marker='o', markersize=3, \n",
    "            label='SLATM_2 \\n $r^2$: {:.2f}, MAE: {:.2f}'.format(\n",
    "             np.corrcoef(slatm_2_df.R_preds, slatm_2_df.Eafw)[0,1] ** 2,\n",
    "             np.mean(np.abs(- dfRS.groupby('mol').dETS.mean() - slatm_2_dde_preds_mean))   \n",
    "            ))\n",
    "\n",
    "\n",
    "ax.errorbar(- dfRS.groupby('mol').dETS.mean(), slatm_diff_dde_preds_mean, \n",
    "            slatm_diff_dde_preds_std, linestyle='', marker='o', markersize=3, \n",
    "            label='SLATM_DIFF \\n $r^2$: {:.2f}, MAE: {:.2f}'.format(\n",
    "             np.corrcoef(slatm_diff_df.R_preds, slatm_diff_df.Eafw)[0,1] ** 2,\n",
    "             np.mean(np.abs(- dfRS.groupby('mol').dETS.mean() - slatm_diff_dde_preds_mean))   \n",
    "            ))\n",
    "\n",
    "\n",
    "ax.errorbar(- dfRS.groupby('mol').dETS.mean(), slatm_diffp_dde_preds_mean, \n",
    "            slatm_diffp_dde_preds_std, linestyle='', marker='o', markersize=3, \n",
    "            label='SLATM_DIFF+ \\n $r^2$: {:.2f}, MAE: {:.2f}'.format(\n",
    "            np.corrcoef(- dfRS.groupby('mol').dETS.mean(), slatm_diffp_dde_preds_mean)[0,1] ** 2,\n",
    "             np.mean(np.abs(- dfRS.groupby('mol').dETS.mean() - slatm_diffp_dde_preds_mean))   \n",
    "            ))\n",
    "\n",
    "ax.plot(-dfRS.groupby('mol').dETS.mean(), -dfRS.groupby('mol').dETS.mean(), c='black', \n",
    "        alpha=0.3, label='x=y')\n",
    "\n",
    "\n",
    "ax.set_xlabel('DFT $\\Delta\\Delta$E [kcal/mol]')\n",
    "\n",
    "ax.set_ylabel('Predicted $\\Delta\\Delta$E [kcal/mol]')\n",
    "\n",
    "ax.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}