03_FS.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from __future__ import print_function\n",
    "\n",
    "from sklearn import datasets\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.svm import SVC\n",
    "\n",
    "import pickle\n",
    "\n",
    "import numpy as np\n",
    "\n",
    "import pandas as pd\n",
    "\n",
    "% matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "# sns.set()\n",
    "\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.feature_selection import RFECV\n",
    "\n",
    "import joblib\n",
    "\n",
    "from libs.container import Container\n",
    "from libs.display import d\n",
    "from libs.experiment import KFoldExperiment, WithAnotherExperiment, roc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "cpu = joblib.cpu_count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'b261': 1, 'b278': 0}"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample = pd.read_pickle(\"data/scaled/sample.pkl.bz2\")\n",
    "sample[\"tile\"] = sample[\"id\"].apply(lambda i: \"b\" + str(i)[1:4])\n",
    "\n",
    "no_features = [\"id\", \"vs_catalog\", \"vs_type\", \"ra_k\", \"dec_k\", \"tile\", \"cls\"] \n",
    "X_columns = [c for c in sample.columns if c not in no_features]\n",
    "\n",
    "grouped = sample.groupby(\"tile\")\n",
    "data = Container({k: grouped.get_group(k).copy() for k in grouped.groups.keys()})\n",
    "data = Container(data=pd.concat([data.b278, data.b261]))\n",
    "\n",
    "cls = {name: idx for idx, name in enumerate(data.data.tile.unique())}\n",
    "data.data[\"cls\"] = data.data.tile.apply(cls.get)\n",
    "\n",
    "del grouped, sample\n",
    "\n",
    "cls"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "cls = {v: v for v in cls.values()}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "RF_PARAMS = {\n",
    "    'max_features': None, 'min_samples_split': 10, 'n_jobs': cpu, \n",
    "    'criterion': 'entropy', 'n_estimators': 500}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = data.data[X_columns].values\n",
    "y = data.data.cls.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 57min 13s, sys: 23.6 s, total: 57min 36s\n",
      "Wall time: 15min 6s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "clf = RandomForestClassifier(**RF_PARAMS)\n",
    "sel = RFECV(clf, n_jobs=cpu, cv=10)\n",
    "sel.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.DataFrame({\n",
    "    \"Feature\": X_columns, \n",
    "    \"Selected\": sel.support_, \"Ranking\": sel.ranking_})\n",
    "df = df.reset_index(drop=True)\n",
    "df.to_pickle(\"data/ranking.pkl.bz2\", compression=\"bz2\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Selected Features:  15\n"
     ]
    }
   ],
   "source": [
    "print(\"Selected Features: \", sum(df.Selected.values))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style  type=\"text/css\" >\n",
       "    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow0_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow0_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow0_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow1_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow1_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow1_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow2_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow2_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow2_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow3_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow3_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow3_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow4_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow4_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow4_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow5_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow5_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow5_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow6_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow6_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow6_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow7_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow7_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow7_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow8_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow8_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow8_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow9_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow9_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow9_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow10_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow10_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow10_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow11_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow11_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow11_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow12_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow12_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow12_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow13_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow13_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow13_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow14_col0 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow14_col1 {\n",
       "            background-color:  #7EA489;\n",
       "        }    #T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow14_col2 {\n",
       "            background-color:  #7EA489;\n",
       "        }</style>  \n",
       "<table id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" > \n",
       "<thead>    <tr> \n",
       "        <th class=\"blank level0\" ></th> \n",
       "        <th class=\"col_heading level0 col0\" >Feature</th> \n",
       "        <th class=\"col_heading level0 col1\" >Ranking</th> \n",
       "        <th class=\"col_heading level0 col2\" >Selected</th> \n",
       "    </tr></thead> \n",
       "<tbody>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row0\" >28</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow0_col0\" class=\"data row0 col0\" >Meanvariance</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow0_col1\" class=\"data row0 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow0_col2\" class=\"data row0 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row1\" >27</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow1_col0\" class=\"data row1 col0\" >Mean</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow1_col1\" class=\"data row1 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow1_col2\" class=\"data row1 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row2\" >17</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow2_col0\" class=\"data row2 col0\" >Freq1_harmonics_amplitude_0</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow2_col1\" class=\"data row2 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow2_col2\" class=\"data row2 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row3\" >37</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow3_col0\" class=\"data row3 col0\" >Psi_eta</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow3_col1\" class=\"data row3 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow3_col2\" class=\"data row3 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row4\" >26</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow4_col0\" class=\"data row4 col0\" >MaxSlope</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow4_col1\" class=\"data row4 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow4_col2\" class=\"data row4 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row5\" >11</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow5_col0\" class=\"data row5 col0\" >Eta_e</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow5_col1\" class=\"data row5 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow5_col2\" class=\"data row5 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row6\" >39</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow6_col0\" class=\"data row6 col0\" >Rcs</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow6_col1\" class=\"data row6 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow6_col2\" class=\"data row6 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row7\" >40</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow7_col0\" class=\"data row7 col0\" >Skew</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow7_col1\" class=\"data row7 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow7_col2\" class=\"data row7 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row8\" >25</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow8_col0\" class=\"data row8 col0\" >LinearTrend</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow8_col1\" class=\"data row8 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow8_col2\" class=\"data row8 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row9\" >6</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow9_col0\" class=\"data row9 col0\" >Beyond1Std</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow9_col1\" class=\"data row9 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow9_col2\" class=\"data row9 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row10\" >49</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow10_col0\" class=\"data row10 col0\" >cnt</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow10_col1\" class=\"data row10 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow10_col2\" class=\"data row10 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row11\" >50</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow11_col0\" class=\"data row11 col0\" >n09_c3</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow11_col1\" class=\"data row11 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow11_col2\" class=\"data row11 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row12\" >54</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow12_col0\" class=\"data row12 col0\" >n09_m2</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow12_col1\" class=\"data row12 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow12_col2\" class=\"data row12 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row13\" >51</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow13_col0\" class=\"data row13 col0\" >n09_hk_color</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow13_col1\" class=\"data row13 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow13_col2\" class=\"data row13 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row14\" >47</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow14_col0\" class=\"data row14 col0\" >c89_m2</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow14_col1\" class=\"data row14 col1\" >1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow14_col2\" class=\"data row14 col2\" >True</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row15\" >12</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow15_col0\" class=\"data row15 col0\" >FluxPercentileRatioMid20</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow15_col1\" class=\"data row15 col1\" >2</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow15_col2\" class=\"data row15 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row16\" >34</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow16_col0\" class=\"data row16 col0\" >PeriodLS</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow16_col1\" class=\"data row16 col1\" >3</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow16_col2\" class=\"data row16 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row17\" >38</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow17_col0\" class=\"data row17 col0\" >Q31</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow17_col1\" class=\"data row17 col1\" >4</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow17_col2\" class=\"data row17 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row18\" >7</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow18_col0\" class=\"data row18 col0\" >CAR_mean</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow18_col1\" class=\"data row18 col1\" >5</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow18_col2\" class=\"data row18 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row19\" >22</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow19_col0\" class=\"data row19 col0\" >Freq1_harmonics_rel_phase_1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow19_col1\" class=\"data row19 col1\" >6</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow19_col2\" class=\"data row19 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row20\" >36</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow20_col0\" class=\"data row20 col0\" >Psi_CS</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow20_col1\" class=\"data row20 col1\" >7</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow20_col2\" class=\"data row20 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row21\" >55</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow21_col0\" class=\"data row21 col0\" >n09_m4</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow21_col1\" class=\"data row21 col1\" >8</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow21_col2\" class=\"data row21 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row22\" >18</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow22_col0\" class=\"data row22 col0\" >Freq1_harmonics_amplitude_1</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow22_col1\" class=\"data row22 col1\" >9</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow22_col2\" class=\"data row22 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row23\" >23</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow23_col0\" class=\"data row23 col0\" >Freq1_harmonics_rel_phase_2</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow23_col1\" class=\"data row23 col1\" >10</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow23_col2\" class=\"data row23 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row24\" >14</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow24_col0\" class=\"data row24 col0\" >FluxPercentileRatioMid50</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow24_col1\" class=\"data row24 col1\" >11</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow24_col2\" class=\"data row24 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row25\" >52</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow25_col0\" class=\"data row25 col0\" >n09_jh_color</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow25_col1\" class=\"data row25 col1\" >12</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow25_col2\" class=\"data row25 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row26\" >56</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow26_col0\" class=\"data row26 col0\" >ppmb</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow26_col1\" class=\"data row26 col1\" >13</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow26_col2\" class=\"data row26 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row27\" >1</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow27_col0\" class=\"data row27 col0\" >AmplitudeH</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow27_col1\" class=\"data row27 col1\" >14</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow27_col2\" class=\"data row27 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row28\" >19</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow28_col0\" class=\"data row28 col0\" >Freq1_harmonics_amplitude_2</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow28_col1\" class=\"data row28 col1\" >15</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow28_col2\" class=\"data row28 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row29\" >48</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow29_col0\" class=\"data row29 col0\" >c89_m4</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow29_col1\" class=\"data row29 col1\" >16</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow29_col2\" class=\"data row29 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row30\" >24</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow30_col0\" class=\"data row30 col0\" >Freq1_harmonics_rel_phase_3</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow30_col1\" class=\"data row30 col1\" >17</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow30_col2\" class=\"data row30 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row31\" >15</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow31_col0\" class=\"data row31 col0\" >FluxPercentileRatioMid65</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow31_col1\" class=\"data row31 col1\" >18</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow31_col2\" class=\"data row31 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row32\" >41</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow32_col0\" class=\"data row32 col0\" >SmallKurtosis</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow32_col1\" class=\"data row32 col1\" >19</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow32_col2\" class=\"data row32 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row33\" >33</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow33_col0\" class=\"data row33 col0\" >PercentDifferenceFluxPercentile</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow33_col1\" class=\"data row33 col1\" >20</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow33_col2\" class=\"data row33 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row34\" >32</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow34_col0\" class=\"data row34 col0\" >PercentAmplitude</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow34_col1\" class=\"data row34 col1\" >21</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow34_col2\" class=\"data row34 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row35\" >20</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow35_col0\" class=\"data row35 col0\" >Freq1_harmonics_amplitude_3</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow35_col1\" class=\"data row35 col1\" >22</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow35_col2\" class=\"data row35 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row36\" >13</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow36_col0\" class=\"data row36 col0\" >FluxPercentileRatioMid35</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow36_col1\" class=\"data row36 col1\" >23</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow36_col2\" class=\"data row36 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row37\" >44</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow37_col0\" class=\"data row37 col0\" >c89_hk_color</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow37_col1\" class=\"data row37 col1\" >24</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow37_col2\" class=\"data row37 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row38\" >16</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow38_col0\" class=\"data row38 col0\" >FluxPercentileRatioMid80</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow38_col1\" class=\"data row38 col1\" >25</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow38_col2\" class=\"data row38 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row39\" >43</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow39_col0\" class=\"data row39 col0\" >c89_c3</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow39_col1\" class=\"data row39 col1\" >26</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow39_col2\" class=\"data row39 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row40\" >30</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow40_col0\" class=\"data row40 col0\" >MedianBRP</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow40_col1\" class=\"data row40 col1\" >27</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow40_col2\" class=\"data row40 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row41\" >46</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow41_col0\" class=\"data row41 col0\" >c89_jk_color</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow41_col1\" class=\"data row41 col1\" >28</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow41_col2\" class=\"data row41 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row42\" >42</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow42_col0\" class=\"data row42 col0\" >Std</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow42_col1\" class=\"data row42 col1\" >29</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow42_col2\" class=\"data row42 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row43\" >8</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow43_col0\" class=\"data row43 col0\" >CAR_sigma</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow43_col1\" class=\"data row43 col1\" >30</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow43_col2\" class=\"data row43 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row44\" >29</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow44_col0\" class=\"data row44 col0\" >MedianAbsDev</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow44_col1\" class=\"data row44 col1\" >31</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow44_col2\" class=\"data row44 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row45\" >31</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow45_col0\" class=\"data row45 col0\" >PairSlopeTrend</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow45_col1\" class=\"data row45 col1\" >32</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow45_col2\" class=\"data row45 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row46\" >4</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow46_col0\" class=\"data row46 col0\" >AmplitudeJK</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow46_col1\" class=\"data row46 col1\" >33</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow46_col2\" class=\"data row46 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row47\" >53</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow47_col0\" class=\"data row47 col0\" >n09_jk_color</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow47_col1\" class=\"data row47 col1\" >34</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow47_col2\" class=\"data row47 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row48\" >45</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow48_col0\" class=\"data row48 col0\" >c89_jh_color</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow48_col1\" class=\"data row48 col1\" >35</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow48_col2\" class=\"data row48 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row49\" >0</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow49_col0\" class=\"data row49 col0\" >Amplitude</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow49_col1\" class=\"data row49 col1\" >36</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow49_col2\" class=\"data row49 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row50\" >3</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow50_col0\" class=\"data row50 col0\" >AmplitudeJH</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow50_col1\" class=\"data row50 col1\" >37</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow50_col2\" class=\"data row50 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row51\" >2</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow51_col0\" class=\"data row51 col0\" >AmplitudeJ</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow51_col1\" class=\"data row51 col1\" >38</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow51_col2\" class=\"data row51 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row52\" >35</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow52_col0\" class=\"data row52 col0\" >Period_fit</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow52_col1\" class=\"data row52 col1\" >39</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow52_col2\" class=\"data row52 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row53\" >10</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow53_col0\" class=\"data row53 col0\" >Con</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow53_col1\" class=\"data row53 col1\" >40</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow53_col2\" class=\"data row53 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row54\" >9</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow54_col0\" class=\"data row54 col0\" >CAR_tau</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow54_col1\" class=\"data row54 col1\" >41</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow54_col2\" class=\"data row54 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row55\" >5</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow55_col0\" class=\"data row55 col0\" >Autocor_length</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow55_col1\" class=\"data row55 col1\" >42</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow55_col2\" class=\"data row55 col2\" >False</td> \n",
       "    </tr>    <tr> \n",
       "        <th id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413c\" class=\"row_heading level0 row56\" >21</th> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow56_col0\" class=\"data row56 col0\" >Freq1_harmonics_rel_phase_0</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow56_col1\" class=\"data row56 col1\" >43</td> \n",
       "        <td id=\"T_c2fcabf4_0861_11e9_9a6b_b16e2597413crow56_col2\" class=\"data row56 col2\" >False</td> \n",
       "    </tr></tbody> \n",
       "</table> "
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f33b57093d0>"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def is_selected(row):\n",
    "    if row.Selected:\n",
    "        return pd.Series('background-color: #7EA489', row.index)\n",
    "    return pd.Series('', row.index)\n",
    "\n",
    "\n",
    "with pd.option_context('display.precision', 4):\n",
    "    html = df.sort_values(\"Ranking\").style.apply(is_selected, axis=1)\n",
    "html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.98795180722891562"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sel.score(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.xlabel(\"Cantidad de features seleccionados\")\n",
    "plt.ylabel(\"Puntaje\")\n",
    "plt.plot(range(1, len(sel.grid_scores_) + 1), sel.grid_scores_)\n",
    "plt.axvline(15, c=\"r\", alpha=0.5, ls=\"--\", label=u\"Features óptimos(15)\")\n",
    "plt.legend()\n",
    "plt.tight_layout()\n",
    "plt.savefig(\"plots/rfe.pdf\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "sX_columns = df[df.Selected].Feature.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "         0.0       0.86      0.90      0.88       994\n",
      "         1.0       0.89      0.85      0.87       998\n",
      "\n",
      "   micro avg       0.87      0.87      0.87      1992\n",
      "   macro avg       0.87      0.87      0.87      1992\n",
      "weighted avg       0.87      0.87      0.87      1992\n",
      "\n",
      "--------------------------------------------------------------------------------\n",
      "CPU times: user 5min 19s, sys: 9.92 s, total: 5min 29s\n",
      "Wall time: 36.7 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "rf = KFoldExperiment(\n",
    "    clf=RandomForestClassifier(**RF_PARAMS), clsnum=cls, \n",
    "    data=data, pcls=1, ncls=0, X_columns=sX_columns, y_column=\"cls\")\n",
    "rf = rf(\"data\", nfolds=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Beyond1Std',\n",
       " 'Eta_e',\n",
       " 'Freq1_harmonics_amplitude_0',\n",
       " 'LinearTrend',\n",
       " 'MaxSlope',\n",
       " 'Mean',\n",
       " 'Meanvariance',\n",
       " 'Psi_eta',\n",
       " 'Rcs',\n",
       " 'Skew',\n",
       " 'c89_m2',\n",
       " 'cnt',\n",
       " 'n09_c3',\n",
       " 'n09_hk_color',\n",
       " 'n09_m2']"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(sX_columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf_all = np.load(\"data/results/model_selection.pkl.npy\")[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = roc({\"RFE-RF\": rf, \"RF\": rf_all});\n",
    "ax.get_children()[0].set_color(\"red\")\n",
    "ax.get_children()[1].set_color(\"blue\")\n",
    "plt.legend()\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.savefig(\"plots/rfe_vs_all.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\\begin{tabular}{lrrr}\n",
      "\\toprule\n",
      " Modelo &  Precision &  Recall &    AUC \\\\\n",
      "\\midrule\n",
      "     RF &     0.8876 &  0.8387 & 0.9412 \\\\\n",
      " RFE-RF &     0.8908 &  0.8497 & 0.9473 \\\\\n",
      "\\bottomrule\n",
      "\\end{tabular}\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\n",
    "    pd.DataFrame([\n",
    "        [\"RF\", rf_all.prec_rec[0], rf_all.prec_rec[1], rf_all.roc_auc],\n",
    "        [\"RFE-RF\", rf.prec_rec[0], rf.prec_rec[1], rf.roc_auc]],\n",
    "        columns=\"Modelo Precision Recall AUC\".split()\n",
    "    ).to_latex(index=False, float_format=\"%.4f\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}