PROJECT_Bioconcentration_factor_06_RF_models_FINALoob_coments_added.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>Name</th>\n",
       "      <th>nAcid</th>\n",
       "      <th>ALogP</th>\n",
       "      <th>ALogp2</th>\n",
       "      <th>AMR</th>\n",
       "      <th>apol</th>\n",
       "      <th>naAromAtom</th>\n",
       "      <th>nAromBond</th>\n",
       "      <th>nAtom</th>\n",
       "      <th>nHeavyAtom</th>\n",
       "      <th>...</th>\n",
       "      <th>WTPT-5</th>\n",
       "      <th>WPATH</th>\n",
       "      <th>WPOL</th>\n",
       "      <th>XLogP</th>\n",
       "      <th>Zagreb</th>\n",
       "      <th>EPI_LogP</th>\n",
       "      <th>XLogP3</th>\n",
       "      <th>VCC_LogP</th>\n",
       "      <th>VCC_LogS</th>\n",
       "      <th>logBCF</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>485</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.3695</td>\n",
       "      <td>0.136530</td>\n",
       "      <td>32.1301</td>\n",
       "      <td>20.748309</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>7</td>\n",
       "      <td>...</td>\n",
       "      <td>3.155330</td>\n",
       "      <td>46</td>\n",
       "      <td>4</td>\n",
       "      <td>0.637</td>\n",
       "      <td>28</td>\n",
       "      <td>-1.44</td>\n",
       "      <td>-4.60</td>\n",
       "      <td>-2.98</td>\n",
       "      <td>-3.71</td>\n",
       "      <td>-2.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>381</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0068</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>40.3611</td>\n",
       "      <td>23.743930</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>10</td>\n",
       "      <td>...</td>\n",
       "      <td>2.836760</td>\n",
       "      <td>122</td>\n",
       "      <td>11</td>\n",
       "      <td>-0.269</td>\n",
       "      <td>42</td>\n",
       "      <td>-0.90</td>\n",
       "      <td>-0.85</td>\n",
       "      <td>-0.66</td>\n",
       "      <td>-0.99</td>\n",
       "      <td>-1.52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>729</td>\n",
       "      <td>0</td>\n",
       "      <td>1.3960</td>\n",
       "      <td>1.948816</td>\n",
       "      <td>29.8218</td>\n",
       "      <td>13.753965</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>10</td>\n",
       "      <td>5</td>\n",
       "      <td>...</td>\n",
       "      <td>2.586760</td>\n",
       "      <td>18</td>\n",
       "      <td>2</td>\n",
       "      <td>0.375</td>\n",
       "      <td>16</td>\n",
       "      <td>0.48</td>\n",
       "      <td>0.72</td>\n",
       "      <td>1.13</td>\n",
       "      <td>-2.41</td>\n",
       "      <td>-1.52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>282</td>\n",
       "      <td>0</td>\n",
       "      <td>-5.2174</td>\n",
       "      <td>27.221263</td>\n",
       "      <td>79.1138</td>\n",
       "      <td>66.718513</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>66</td>\n",
       "      <td>25</td>\n",
       "      <td>...</td>\n",
       "      <td>18.287187</td>\n",
       "      <td>2556</td>\n",
       "      <td>22</td>\n",
       "      <td>2.856</td>\n",
       "      <td>98</td>\n",
       "      <td>2.87</td>\n",
       "      <td>1.73</td>\n",
       "      <td>2.35</td>\n",
       "      <td>-4.55</td>\n",
       "      <td>-1.40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>333</td>\n",
       "      <td>0</td>\n",
       "      <td>0.2553</td>\n",
       "      <td>0.065178</td>\n",
       "      <td>50.1468</td>\n",
       "      <td>29.963102</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>28</td>\n",
       "      <td>14</td>\n",
       "      <td>...</td>\n",
       "      <td>5.609667</td>\n",
       "      <td>337</td>\n",
       "      <td>18</td>\n",
       "      <td>0.463</td>\n",
       "      <td>64</td>\n",
       "      <td>-0.67</td>\n",
       "      <td>-0.57</td>\n",
       "      <td>0.01</td>\n",
       "      <td>-2.33</td>\n",
       "      <td>-1.16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1002</td>\n",
       "      <td>988</td>\n",
       "      <td>0</td>\n",
       "      <td>3.7215</td>\n",
       "      <td>13.849562</td>\n",
       "      <td>28.0780</td>\n",
       "      <td>35.353965</td>\n",
       "      <td>12</td>\n",
       "      <td>13</td>\n",
       "      <td>22</td>\n",
       "      <td>17</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>470</td>\n",
       "      <td>30</td>\n",
       "      <td>6.295</td>\n",
       "      <td>88</td>\n",
       "      <td>6.98</td>\n",
       "      <td>6.75</td>\n",
       "      <td>6.79</td>\n",
       "      <td>-7.58</td>\n",
       "      <td>4.85</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1003</td>\n",
       "      <td>430</td>\n",
       "      <td>0</td>\n",
       "      <td>5.2002</td>\n",
       "      <td>27.042080</td>\n",
       "      <td>48.6066</td>\n",
       "      <td>46.636551</td>\n",
       "      <td>12</td>\n",
       "      <td>12</td>\n",
       "      <td>34</td>\n",
       "      <td>27</td>\n",
       "      <td>...</td>\n",
       "      <td>9.432636</td>\n",
       "      <td>1630</td>\n",
       "      <td>48</td>\n",
       "      <td>6.917</td>\n",
       "      <td>142</td>\n",
       "      <td>7.68</td>\n",
       "      <td>6.22</td>\n",
       "      <td>5.36</td>\n",
       "      <td>-7.31</td>\n",
       "      <td>5.08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1004</td>\n",
       "      <td>683</td>\n",
       "      <td>0</td>\n",
       "      <td>4.4658</td>\n",
       "      <td>19.943370</td>\n",
       "      <td>33.6936</td>\n",
       "      <td>36.867172</td>\n",
       "      <td>12</td>\n",
       "      <td>13</td>\n",
       "      <td>22</td>\n",
       "      <td>18</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>573</td>\n",
       "      <td>30</td>\n",
       "      <td>5.336</td>\n",
       "      <td>94</td>\n",
       "      <td>7.62</td>\n",
       "      <td>7.12</td>\n",
       "      <td>7.25</td>\n",
       "      <td>-8.21</td>\n",
       "      <td>5.69</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1005</td>\n",
       "      <td>435</td>\n",
       "      <td>0</td>\n",
       "      <td>1.7669</td>\n",
       "      <td>3.121936</td>\n",
       "      <td>117.3685</td>\n",
       "      <td>77.660892</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>72</td>\n",
       "      <td>28</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2214</td>\n",
       "      <td>45</td>\n",
       "      <td>9.896</td>\n",
       "      <td>146</td>\n",
       "      <td>10.24</td>\n",
       "      <td>7.95</td>\n",
       "      <td>7.98</td>\n",
       "      <td>-6.01</td>\n",
       "      <td>6.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1006</td>\n",
       "      <td>747</td>\n",
       "      <td>0</td>\n",
       "      <td>5.7396</td>\n",
       "      <td>32.943008</td>\n",
       "      <td>94.8412</td>\n",
       "      <td>39.400000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>20</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>650</td>\n",
       "      <td>47</td>\n",
       "      <td>6.094</td>\n",
       "      <td>114</td>\n",
       "      <td>8.39</td>\n",
       "      <td>6.27</td>\n",
       "      <td>7.08</td>\n",
       "      <td>-8.94</td>\n",
       "      <td>6.43</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1007 rows × 553 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Name  nAcid   ALogP     ALogp2       AMR       apol  naAromAtom  \\\n",
       "0      485      0 -0.3695   0.136530   32.1301  20.748309           0   \n",
       "1      381      0  0.0068   0.000046   40.3611  23.743930           0   \n",
       "2      729      0  1.3960   1.948816   29.8218  13.753965           0   \n",
       "3      282      0 -5.2174  27.221263   79.1138  66.718513           0   \n",
       "4      333      0  0.2553   0.065178   50.1468  29.963102           0   \n",
       "...    ...    ...     ...        ...       ...        ...         ...   \n",
       "1002   988      0  3.7215  13.849562   28.0780  35.353965          12   \n",
       "1003   430      0  5.2002  27.042080   48.6066  46.636551          12   \n",
       "1004   683      0  4.4658  19.943370   33.6936  36.867172          12   \n",
       "1005   435      0  1.7669   3.121936  117.3685  77.660892           0   \n",
       "1006   747      0  5.7396  32.943008   94.8412  39.400000           0   \n",
       "\n",
       "      nAromBond  nAtom  nHeavyAtom  ...     WTPT-5  WPATH  WPOL  XLogP  \\\n",
       "0             0     20           7  ...   3.155330     46     4  0.637   \n",
       "1             0     20          10  ...   2.836760    122    11 -0.269   \n",
       "2             0     10           5  ...   2.586760     18     2  0.375   \n",
       "3             0     66          25  ...  18.287187   2556    22  2.856   \n",
       "4             0     28          14  ...   5.609667    337    18  0.463   \n",
       "...         ...    ...         ...  ...        ...    ...   ...    ...   \n",
       "1002         13     22          17  ...   0.000000    470    30  6.295   \n",
       "1003         12     34          27  ...   9.432636   1630    48  6.917   \n",
       "1004         13     22          18  ...   0.000000    573    30  5.336   \n",
       "1005          0     72          28  ...   0.000000   2214    45  9.896   \n",
       "1006          0     20          20  ...   0.000000    650    47  6.094   \n",
       "\n",
       "      Zagreb  EPI_LogP  XLogP3  VCC_LogP  VCC_LogS  logBCF  \n",
       "0         28     -1.44   -4.60     -2.98     -3.71   -2.00  \n",
       "1         42     -0.90   -0.85     -0.66     -0.99   -1.52  \n",
       "2         16      0.48    0.72      1.13     -2.41   -1.52  \n",
       "3         98      2.87    1.73      2.35     -4.55   -1.40  \n",
       "4         64     -0.67   -0.57      0.01     -2.33   -1.16  \n",
       "...      ...       ...     ...       ...       ...     ...  \n",
       "1002      88      6.98    6.75      6.79     -7.58    4.85  \n",
       "1003     142      7.68    6.22      5.36     -7.31    5.08  \n",
       "1004      94      7.62    7.12      7.25     -8.21    5.69  \n",
       "1005     146     10.24    7.95      7.98     -6.01    6.00  \n",
       "1006     114      8.39    6.27      7.08     -8.94    6.43  \n",
       "\n",
       "[1007 rows x 553 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "mergedDescs=pd.read_csv(\"merged_Descs_sorted.csv\")\n",
    "data=mergedDescs.copy()\n",
    "# made a copy of original dataset\n",
    "data\n",
    "# this is the dataset which was sorted according to 'logBCF'in Excel by Aleksandra\n",
    "# so when there were 2 equal values, Excel sorted those two in ascending order (eg. 333 and 367)\n",
    "# but in principle, it does not matter anymore, you could take the original dataset as well"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# loaded datasets with IDs or Names of the Test-train split as provided by authors of the reference publication\n",
    "test_names=pd.read_csv(\"ExternalID.csv\", header=None)\n",
    "train_names=pd.read_csv(\"TrainID.csv\", header=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# useful when need to see more\n",
    "# data = pd.set_option('display.max_rows', 1007)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# added new column 'BCF' with 0 or 1 according to the threshold 3.0\n",
    "data[\"BCF\"] = data.apply(lambda row: 0 if row[\"logBCF\"] < 3.0 else 1, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# removed original logBCF column because now we have binary BCF column\n",
    "data = data.drop(['logBCF'], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    798\n",
       "1    209\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check what is the distribution of our labels\n",
    "data.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "# for splitting the dataset, selected rows based on unique values in the 'Name' column using or (|)\n",
    "# unique values were chosen from original datasets provided by authors of publication\n",
    "test= data[(data['Name']==729)|\n",
    "           (data['Name']==367)|\n",
    "           (data['Name']==475)|\n",
    "           (data['Name']==951)|\n",
    "           (data['Name']==21)|\n",
    "           (data['Name']==810)|\n",
    "           (data['Name']==1009)|\n",
    "           (data['Name']==273)|\n",
    "           (data['Name']==168)|\n",
    "           (data['Name']==586)|\n",
    "           (data['Name']==594)|\n",
    "           (data['Name']==22)|\n",
    "           (data['Name']==13)|\n",
    "           (data['Name']==941)|\n",
    "           (data['Name']==4)|\n",
    "           (data['Name']==253)|\n",
    "           (data['Name']==748)|\n",
    "           (data['Name']==807)|\n",
    "           (data['Name']==23)|\n",
    "           (data['Name']==953)|\n",
    "           (data['Name']==876)|\n",
    "           (data['Name']==571)|\n",
    "           (data['Name']==600)|\n",
    "           (data['Name']==372)|\n",
    "           (data['Name']==492)|\n",
    "           (data['Name']==802)|\n",
    "           (data['Name']==3)|\n",
    "           (data['Name']==118)|\n",
    "           (data['Name']==606)|\n",
    "           (data['Name']==42)|\n",
    "           (data['Name']==569)|\n",
    "           (data['Name']==505)|\n",
    "           (data['Name']==324)|\n",
    "           (data['Name']==588)|\n",
    "           (data['Name']==173)|\n",
    "           (data['Name']==65)|\n",
    "           (data['Name']==867)|\n",
    "           (data['Name']==145)|\n",
    "           (data['Name']==1032)|\n",
    "           (data['Name']==98)|\n",
    "           (data['Name']==101)|\n",
    "           (data['Name']==478)|\n",
    "           (data['Name']==30)|\n",
    "           (data['Name']==959)|\n",
    "           (data['Name']==960)|\n",
    "           (data['Name']==396)|\n",
    "           (data['Name']==621)|\n",
    "           (data['Name']==860)|\n",
    "           (data['Name']==559)|\n",
    "           (data['Name']==100)|\n",
    "           (data['Name']==568)|\n",
    "           (data['Name']==847)|\n",
    "           (data['Name']==194)|\n",
    "           (data['Name']==120)|\n",
    "           (data['Name']==24)|\n",
    "           (data['Name']==970)|\n",
    "           (data['Name']==149)|\n",
    "           (data['Name']==99)|\n",
    "           (data['Name']==204)|\n",
    "           (data['Name']==183)|\n",
    "           (data['Name']==10)|\n",
    "           (data['Name']==652)|\n",
    "           (data['Name']==840)|\n",
    "           (data['Name']==774)|\n",
    "           (data['Name']==576)|\n",
    "           (data['Name']==496)|\n",
    "           (data['Name']==74)|\n",
    "           (data['Name']==557)|\n",
    "           (data['Name']==871)|\n",
    "           (data['Name']==1030)|\n",
    "           (data['Name']==518)|\n",
    "           (data['Name']==126)|\n",
    "           (data['Name']==964)|\n",
    "           (data['Name']==215)|\n",
    "           (data['Name']==73)|\n",
    "           (data['Name']==555)|\n",
    "           (data['Name']==412)|\n",
    "           (data['Name']==150)|\n",
    "           (data['Name']==719)|\n",
    "           (data['Name']==537)|\n",
    "           (data['Name']==612)|\n",
    "           (data['Name']==136)|\n",
    "           (data['Name']==75)|\n",
    "           (data['Name']==26)|\n",
    "           (data['Name']==707)|\n",
    "           (data['Name']==424)|\n",
    "           (data['Name']==917)|\n",
    "           (data['Name']==565)|\n",
    "           (data['Name']==79)|\n",
    "           (data['Name']==66)|\n",
    "           (data['Name']==248)|\n",
    "           (data['Name']==155)|\n",
    "           (data['Name']==237)|\n",
    "           (data['Name']==721)|\n",
    "           (data['Name']==212)|\n",
    "           (data['Name']==523)|\n",
    "           (data['Name']==541)|\n",
    "           (data['Name']==112)|\n",
    "           (data['Name']==668)|\n",
    "           (data['Name']==222)|\n",
    "           (data['Name']==940)|\n",
    "           (data['Name']==491)|\n",
    "           (data['Name']==924)|\n",
    "           (data['Name']==146)|\n",
    "           (data['Name']==64)|\n",
    "           (data['Name']==539)|\n",
    "           (data['Name']==190)|\n",
    "           (data['Name']==1007)|\n",
    "           (data['Name']==511)|\n",
    "           (data['Name']==854)|\n",
    "           (data['Name']==334)|\n",
    "           (data['Name']==617)|\n",
    "           (data['Name']==409)|\n",
    "           (data['Name']==582)|\n",
    "           (data['Name']==61)|\n",
    "           (data['Name']==156)|\n",
    "           (data['Name']==1036)|\n",
    "           (data['Name']==543)|\n",
    "           (data['Name']==706)|\n",
    "           (data['Name']==95)|\n",
    "           (data['Name']==442)|\n",
    "           (data['Name']==649)|\n",
    "           (data['Name']==989)|\n",
    "           (data['Name']==121)|\n",
    "           (data['Name']==388)|\n",
    "           (data['Name']==708)|\n",
    "           (data['Name']==616)|\n",
    "           (data['Name']==436)|\n",
    "           (data['Name']==444)|\n",
    "           (data['Name']==968)|\n",
    "           (data['Name']==734)|\n",
    "           (data['Name']==205)|\n",
    "           (data['Name']==611)|\n",
    "           (data['Name']==610)|\n",
    "           (data['Name']==92)|\n",
    "           (data['Name']==819)|\n",
    "           (data['Name']==535)|\n",
    "           (data['Name']==653)|\n",
    "           (data['Name']==1033)|\n",
    "           (data['Name']==943)|\n",
    "           (data['Name']==656)|\n",
    "           (data['Name']==288)|\n",
    "           (data['Name']==347)|\n",
    "           (data['Name']==698)|\n",
    "           (data['Name']==227)|\n",
    "           (data['Name']==337)|\n",
    "           (data['Name']==745)|\n",
    "           (data['Name']==293)|\n",
    "           (data['Name']==452)|\n",
    "           (data['Name']==609)|\n",
    "           (data['Name']==384)|\n",
    "           (data['Name']==763)|\n",
    "           (data['Name']==363)|\n",
    "           (data['Name']==368)|\n",
    "           (data['Name']==1006)|\n",
    "           (data['Name']==546)|\n",
    "           (data['Name']==137)|\n",
    "           (data['Name']==797)|\n",
    "           (data['Name']==762)|\n",
    "           (data['Name']==1018)|\n",
    "           (data['Name']==153)|\n",
    "           (data['Name']==769)|\n",
    "           (data['Name']==312)|\n",
    "           (data['Name']==1020)|\n",
    "           (data['Name']==216)|\n",
    "           (data['Name']==151)|\n",
    "           (data['Name']==815)|\n",
    "           (data['Name']==800)|\n",
    "           (data['Name']==290)|\n",
    "           (data['Name']==345)|\n",
    "           (data['Name']==371)|\n",
    "           (data['Name']==401)|\n",
    "           (data['Name']==733)|\n",
    "           (data['Name']==768)|\n",
    "           (data['Name']==40)|\n",
    "           (data['Name']==58)|\n",
    "           (data['Name']==948)|\n",
    "           (data['Name']==132)|\n",
    "           (data['Name']==728)|\n",
    "           (data['Name']==676)|\n",
    "           (data['Name']==855)|\n",
    "           (data['Name']==325)|\n",
    "           (data['Name']==737)|\n",
    "           (data['Name']==309)|\n",
    "           (data['Name']==355)|\n",
    "           (data['Name']==925)|\n",
    "           (data['Name']==8)|\n",
    "           (data['Name']==743)|\n",
    "           (data['Name']==913)|\n",
    "           (data['Name']==782)|\n",
    "           (data['Name']==428)|\n",
    "           (data['Name']==301)|\n",
    "           (data['Name']==431)|\n",
    "           (data['Name']==592)|\n",
    "           (data['Name']==15)|\n",
    "           (data['Name']==903)|\n",
    "           (data['Name']==276)|\n",
    "           (data['Name']==315)|\n",
    "           (data['Name']==302)|\n",
    "           (data['Name']==525)|\n",
    "           (data['Name']==316)|\n",
    "           (data['Name']==35)|\n",
    "           (data['Name']==422)|\n",
    "           (data['Name']==723)|\n",
    "           (data['Name']==59)|\n",
    "           (data['Name']==1021)|\n",
    "           (data['Name']==184)|\n",
    "           (data['Name']==53)|\n",
    "           (data['Name']==892)|\n",
    "           (data['Name']==69)|\n",
    "           (data['Name']==973)|\n",
    "           (data['Name']==322)|\n",
    "           (data['Name']==310)|\n",
    "           (data['Name']==623)|\n",
    "           (data['Name']==965)|\n",
    "           (data['Name']==124)|\n",
    "           (data['Name']==420)|\n",
    "           (data['Name']==16)|\n",
    "           (data['Name']==882)|\n",
    "           (data['Name']==961)|\n",
    "           (data['Name']==90)|\n",
    "           (data['Name']==283)|\n",
    "           (data['Name']==839)|\n",
    "           (data['Name']==39)|\n",
    "           (data['Name']==856)|\n",
    "           (data['Name']==752)|\n",
    "           (data['Name']==1035)|\n",
    "           (data['Name']==540)|\n",
    "           (data['Name']==1004)|\n",
    "           (data['Name']==545)|\n",
    "           (data['Name']==863)|\n",
    "           (data['Name']==296)|\n",
    "           (data['Name']==447)|\n",
    "           (data['Name']==558)|\n",
    "           (data['Name']==286)|\n",
    "           (data['Name']==348)|\n",
    "           (data['Name']==111)|\n",
    "           (data['Name']==831)|\n",
    "           (data['Name']==130)|\n",
    "           (data['Name']==710)|\n",
    "           (data['Name']==56)|\n",
    "           (data['Name']==242)|\n",
    "           (data['Name']==779)|\n",
    "           (data['Name']==643)|\n",
    "           (data['Name']==574)|\n",
    "           (data['Name']==672)|\n",
    "           (data['Name']==356)|\n",
    "           (data['Name']==2)|\n",
    "           (data['Name']==147)|\n",
    "           (data['Name']==317)|\n",
    "           (data['Name']==502)|\n",
    "           (data['Name']==853)|\n",
    "           (data['Name']==932)|\n",
    "           (data['Name']==937)|\n",
    "           (data['Name']==437)|\n",
    "           (data['Name']==143)|\n",
    "           (data['Name']==922)|\n",
    "           (data['Name']==773)|\n",
    "           (data['Name']==659)|\n",
    "           (data['Name']==378)|\n",
    "           (data['Name']==230)|\n",
    "           (data['Name']==241)|\n",
    "           (data['Name']==603)|\n",
    "           (data['Name']==343)|\n",
    "           (data['Name']==524)|\n",
    "           (data['Name']==28)|\n",
    "           (data['Name']==307)|\n",
    "           (data['Name']==259)|\n",
    "           (data['Name']==342)|\n",
    "           (data['Name']==454)|\n",
    "           (data['Name']==791)|\n",
    "           (data['Name']==81)|\n",
    "           (data['Name']==697)|\n",
    "           (data['Name']==209)|\n",
    "           (data['Name']==415)|\n",
    "           (data['Name']==125)|\n",
    "           (data['Name']==599)|\n",
    "           (data['Name']==662)|\n",
    "           (data['Name']==593)|\n",
    "           (data['Name']==313)|\n",
    "           (data['Name']==628)|\n",
    "           (data['Name']==664)|\n",
    "           (data['Name']==705)|\n",
    "           (data['Name']==117)|\n",
    "           (data['Name']==760)|\n",
    "           (data['Name']==878)|\n",
    "           (data['Name']==534)|\n",
    "           (data['Name']==232)|\n",
    "           (data['Name']==234)|\n",
    "           (data['Name']==977)|\n",
    "           (data['Name']==318)|\n",
    "           (data['Name']==688)|\n",
    "           (data['Name']==104)|\n",
    "           (data['Name']==969)|\n",
    "           (data['Name']==292)|\n",
    "           (data['Name']==445)|\n",
    "           (data['Name']==482)|\n",
    "           (data['Name']==678)|\n",
    "           (data['Name']==618)|\n",
    "           (data['Name']==266)|\n",
    "           (data['Name']==642)|\n",
    "           (data['Name']==140)|\n",
    "           (data['Name']==249)|\n",
    "           (data['Name']==873)|\n",
    "           (data['Name']==984)|\n",
    "           (data['Name']==421)|\n",
    "           (data['Name']==279)|\n",
    "           (data['Name']==185)|\n",
    "           (data['Name']==758)|\n",
    "           (data['Name']==201)|\n",
    "           (data['Name']==682)|\n",
    "           (data['Name']==211)|\n",
    "           (data['Name']==1029)|\n",
    "           (data['Name']==981)|\n",
    "           (data['Name']==887)|\n",
    "           (data['Name']==133)|\n",
    "           (data['Name']==391)|\n",
    "           (data['Name']==684)|\n",
    "           (data['Name']==670)|\n",
    "           (data['Name']==694)|\n",
    "           (data['Name']==641)|\n",
    "           (data['Name']==247)|\n",
    "           (data['Name']==256)|\n",
    "           (data['Name']==686)|\n",
    "           (data['Name']==218)|\n",
    "           (data['Name']==893)|\n",
    "           (data['Name']==875)|\n",
    "           (data['Name']==18)|\n",
    "           (data['Name']==679)|\n",
    "           (data['Name']==280)|\n",
    "           (data['Name']==999)|\n",
    "           (data['Name']==323)|\n",
    "           (data['Name']==687)|\n",
    "           (data['Name']==683)\n",
    "          ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# the same approach as for the test set above\n",
    "train=data[(data['Name']==485)|\n",
    "           (data['Name']==381)|\n",
    "        (data['Name']==282)|\n",
    "        (data['Name']==333)|\n",
    "        (data['Name']==275)|\n",
    "        (data['Name']==494)|\n",
    "        (data['Name']==161)|\n",
    "        (data['Name']==584)|\n",
    "        (data['Name']==823)|\n",
    "        (data['Name']==410)|\n",
    "        (data['Name']==352)|\n",
    "        (data['Name']==908)|\n",
    "        (data['Name']==261)|\n",
    "        (data['Name']==715)|\n",
    "        (data['Name']==551)|\n",
    "        (data['Name']==597)|\n",
    "        (data['Name']==711)|\n",
    "        (data['Name']==716)|\n",
    "        (data['Name']==846)|\n",
    "        (data['Name']==103)|\n",
    "        (data['Name']==224)|\n",
    "        (data['Name']==54)|\n",
    "        (data['Name']==792)|\n",
    "        (data['Name']==1)|\n",
    "        (data['Name']==457)|\n",
    "        (data['Name']==730)|\n",
    "        (data['Name']==829)|\n",
    "        (data['Name']==869)|\n",
    "        (data['Name']==219)|\n",
    "        (data['Name']==601)|\n",
    "        (data['Name']==105)|\n",
    "        (data['Name']==37)|\n",
    "        (data['Name']==148)|\n",
    "        (data['Name']==532)|\n",
    "        (data['Name']==416)|\n",
    "        (data['Name']==509)|\n",
    "        (data['Name']==453)|\n",
    "        (data['Name']==468)|\n",
    "        (data['Name']==179)|\n",
    "        (data['Name']==154)|\n",
    "        (data['Name']==615)|\n",
    "        (data['Name']==771)|\n",
    "        (data['Name']==14)|\n",
    "        (data['Name']==195)|\n",
    "        (data['Name']==265)|\n",
    "        (data['Name']==640)|\n",
    "        (data['Name']==129)|\n",
    "        (data['Name']==335)|\n",
    "        (data['Name']==455)|\n",
    "        (data['Name']==477)|\n",
    "        (data['Name']==498)|\n",
    "        (data['Name']==573)|\n",
    "        (data['Name']==1025)|\n",
    "        (data['Name']==199)|\n",
    "        (data['Name']==134)|\n",
    "        (data['Name']==91)|\n",
    "        (data['Name']==268)|\n",
    "        (data['Name']==197)|\n",
    "        (data['Name']==515)|\n",
    "        (data['Name']==80)|\n",
    "        (data['Name']==508)|\n",
    "        (data['Name']==838)|\n",
    "        (data['Name']==461)|\n",
    "        (data['Name']==722)|\n",
    "        (data['Name']==203)|\n",
    "        (data['Name']==604)|\n",
    "        (data['Name']==41)|\n",
    "        (data['Name']==11)|\n",
    "        (data['Name']==48)|\n",
    "        (data['Name']==872)|\n",
    "        (data['Name']==198)|\n",
    "        (data['Name']==735)|\n",
    "        (data['Name']==229)|\n",
    "        (data['Name']==906)|\n",
    "        (data['Name']==390)|\n",
    "        (data['Name']==581)|\n",
    "        (data['Name']==598)|\n",
    "        (data['Name']==1031)|\n",
    "        (data['Name']==221)|\n",
    "        (data['Name']==107)|\n",
    "        (data['Name']==57)|\n",
    "        (data['Name']==258)|\n",
    "        (data['Name']==171)|\n",
    "        (data['Name']==693)|\n",
    "        (data['Name']==645)|\n",
    "        (data['Name']==31)|\n",
    "        (data['Name']==884)|\n",
    "        (data['Name']==210)|\n",
    "        (data['Name']==563)|\n",
    "        (data['Name']==650)|\n",
    "        (data['Name']==361)|\n",
    "        (data['Name']==364)|\n",
    "        (data['Name']==828)|\n",
    "        (data['Name']==12)|\n",
    "        (data['Name']==564)|\n",
    "        (data['Name']==174)|\n",
    "        (data['Name']==536)|\n",
    "        (data['Name']==97)|\n",
    "        (data['Name']==519)|\n",
    "        (data['Name']==483)|\n",
    "        (data['Name']==963)|\n",
    "        (data['Name']==277)|\n",
    "        (data['Name']==583)|\n",
    "        (data['Name']==552)|\n",
    "        (data['Name']==514)|\n",
    "        (data['Name']==119)|\n",
    "        (data['Name']==304)|\n",
    "        (data['Name']==362)|\n",
    "        (data['Name']==213)|\n",
    "        (data['Name']==71)|\n",
    "        (data['Name']==231)|\n",
    "        (data['Name']==51)|\n",
    "        (data['Name']==942)|\n",
    "        (data['Name']==1019)|\n",
    "        (data['Name']==113)|\n",
    "        (data['Name']==660)|\n",
    "        (data['Name']==510)|\n",
    "        (data['Name']==142)|\n",
    "        (data['Name']==47)|\n",
    "        (data['Name']==966)|\n",
    "        (data['Name']==985)|\n",
    "        (data['Name']==157)|\n",
    "        (data['Name']==17)|\n",
    "        (data['Name']==46)|\n",
    "        (data['Name']==799)|\n",
    "        (data['Name']==544)|\n",
    "        (data['Name']==139)|\n",
    "        (data['Name']==400)|\n",
    "        (data['Name']==62)|\n",
    "        (data['Name']==252)|\n",
    "        (data['Name']==786)|\n",
    "        (data['Name']==820)|\n",
    "        (data['Name']==251)|\n",
    "        (data['Name']==862)|\n",
    "        (data['Name']==935)|\n",
    "        (data['Name']==671)|\n",
    "        (data['Name']==624)|\n",
    "        (data['Name']==633)|\n",
    "        (data['Name']==978)|\n",
    "        (data['Name']==383)|\n",
    "        (data['Name']==516)|\n",
    "        (data['Name']==891)|\n",
    "        (data['Name']==931)|\n",
    "        (data['Name']==178)|\n",
    "        (data['Name']==202)|\n",
    "        (data['Name']==562)|\n",
    "        (data['Name']==625)|\n",
    "        (data['Name']==60)|\n",
    "        (data['Name']==96)|\n",
    "        (data['Name']==920)|\n",
    "        (data['Name']==547)|\n",
    "        (data['Name']==351)|\n",
    "        (data['Name']==635)|\n",
    "        (data['Name']==776)|\n",
    "        (data['Name']==228)|\n",
    "        (data['Name']==45)|\n",
    "        (data['Name']==88)|\n",
    "        (data['Name']==192)|\n",
    "        (data['Name']==44)|\n",
    "        (data['Name']==549)|\n",
    "        (data['Name']==70)|\n",
    "        (data['Name']==499)|\n",
    "        (data['Name']==67)|\n",
    "        (data['Name']==696)|\n",
    "        (data['Name']==602)|\n",
    "        (data['Name']==36)|\n",
    "        (data['Name']==952)|\n",
    "        (data['Name']==975)|\n",
    "        (data['Name']==848)|\n",
    "        (data['Name']==440)|\n",
    "        (data['Name']==1005)|\n",
    "        (data['Name']==94)|\n",
    "        (data['Name']==550)|\n",
    "        (data['Name']==837)|\n",
    "        (data['Name']==1015)|\n",
    "        (data['Name']==507)|\n",
    "        (data['Name']==780)|\n",
    "        (data['Name']==34)|\n",
    "        (data['Name']==527)|\n",
    "        (data['Name']==407)|\n",
    "        (data['Name']==858)|\n",
    "        (data['Name']==577)|\n",
    "        (data['Name']==956)|\n",
    "        (data['Name']==667)|\n",
    "        (data['Name']==1022)|\n",
    "        (data['Name']==162)|\n",
    "        (data['Name']==620)|\n",
    "        (data['Name']==651)|\n",
    "        (data['Name']==830)|\n",
    "        (data['Name']==513)|\n",
    "        (data['Name']==865)|\n",
    "        (data['Name']==188)|\n",
    "        (data['Name']==298)|\n",
    "        (data['Name']==626)|\n",
    "        (data['Name']==548)|\n",
    "        (data['Name']==303)|\n",
    "        (data['Name']==84)|\n",
    "        (data['Name']==947)|\n",
    "        (data['Name']==566)|\n",
    "        (data['Name']==954)|\n",
    "        (data['Name']==506)|\n",
    "        (data['Name']==267)|\n",
    "        (data['Name']==319)|\n",
    "        (data['Name']==353)|\n",
    "        (data['Name']==929)|\n",
    "        (data['Name']==83)|\n",
    "        (data['Name']==833)|\n",
    "        (data['Name']==9)|\n",
    "        (data['Name']==967)|\n",
    "        (data['Name']==880)|\n",
    "        (data['Name']==957)|\n",
    "        (data['Name']==27)|\n",
    "        (data['Name']==822)|\n",
    "        (data['Name']==979)|\n",
    "        (data['Name']==332)|\n",
    "        (data['Name']==1003)|\n",
    "        (data['Name']==556)|\n",
    "        (data['Name']==843)|\n",
    "        (data['Name']==888)|\n",
    "        (data['Name']==5)|\n",
    "        (data['Name']==533)|\n",
    "        (data['Name']==330)|\n",
    "        (data['Name']==781)|\n",
    "        (data['Name']==927)|\n",
    "        (data['Name']==526)|\n",
    "        (data['Name']==25)|\n",
    "        (data['Name']==114)|\n",
    "        (data['Name']==945)|\n",
    "        (data['Name']==834)|\n",
    "        (data['Name']==764)|\n",
    "        (data['Name']==859)|\n",
    "        (data['Name']==223)|\n",
    "        (data['Name']==375)|\n",
    "        (data['Name']==1010)|\n",
    "        (data['Name']==832)|\n",
    "        (data['Name']==425)|\n",
    "        (data['Name']==634)|\n",
    "        (data['Name']==704)|\n",
    "        (data['Name']==844)|\n",
    "        (data['Name']==87)|\n",
    "        (data['Name']==877)|\n",
    "        (data['Name']==20)|\n",
    "        (data['Name']==852)|\n",
    "        (data['Name']==1001)|\n",
    "        (data['Name']==938)|\n",
    "        (data['Name']==495)|\n",
    "        (data['Name']==665)|\n",
    "        (data['Name']==427)|\n",
    "        (data['Name']==845)|\n",
    "        (data['Name']==135)|\n",
    "        (data['Name']==585)|\n",
    "        (data['Name']==365)|\n",
    "        (data['Name']==918)|\n",
    "        (data['Name']==187)|\n",
    "        (data['Name']==169)|\n",
    "        (data['Name']==474)|\n",
    "        (data['Name']==619)|\n",
    "        (data['Name']==939)|\n",
    "        (data['Name']==567)|\n",
    "        (data['Name']==666)|\n",
    "        (data['Name']==864)|\n",
    "        (data['Name']==486)|\n",
    "        (data['Name']==914)|\n",
    "        (data['Name']==622)|\n",
    "        (data['Name']==849)|\n",
    "        (data['Name']==836)|\n",
    "        (data['Name']==614)|\n",
    "        (data['Name']==467)|\n",
    "        (data['Name']==340)|\n",
    "        (data['Name']==759)|\n",
    "        (data['Name']==481)|\n",
    "        (data['Name']==374)|\n",
    "        (data['Name']==1008)|\n",
    "        (data['Name']==77)|\n",
    "        (data['Name']==958)|\n",
    "        (data['Name']==189)|\n",
    "        (data['Name']==528)|\n",
    "        (data['Name']==1012)|\n",
    "        (data['Name']==226)|\n",
    "        (data['Name']==466)|\n",
    "        (data['Name']==1002)|\n",
    "        (data['Name']==419)|\n",
    "        (data['Name']==257)|\n",
    "        (data['Name']==271)|\n",
    "        (data['Name']==373)|\n",
    "        (data['Name']==393)|\n",
    "        (data['Name']==191)|\n",
    "        (data['Name']==572)|\n",
    "        (data['Name']==49)|\n",
    "        (data['Name']==287)|\n",
    "        (data['Name']==857)|\n",
    "        (data['Name']==472)|\n",
    "        (data['Name']==811)|\n",
    "        (data['Name']==790)|\n",
    "        (data['Name']==850)|\n",
    "        (data['Name']==934)|\n",
    "        (data['Name']==661)|\n",
    "        (data['Name']==784)|\n",
    "        (data['Name']==632)|\n",
    "        (data['Name']==955)|\n",
    "        (data['Name']==200)|\n",
    "        (data['Name']==55)|\n",
    "        (data['Name']==736)|\n",
    "        (data['Name']==928)|\n",
    "        (data['Name']==68)|\n",
    "        (data['Name']==43)|\n",
    "        (data['Name']==500)|\n",
    "        (data['Name']==1034)|\n",
    "        (data['Name']==180)|\n",
    "        (data['Name']==439)|\n",
    "        (data['Name']==497)|\n",
    "        (data['Name']==264)|\n",
    "        (data['Name']==450)|\n",
    "        (data['Name']==269)|\n",
    "        (data['Name']==731)|\n",
    "        (data['Name']==911)|\n",
    "        (data['Name']==403)|\n",
    "        (data['Name']==542)|\n",
    "        (data['Name']==1013)|\n",
    "        (data['Name']==673)|\n",
    "        (data['Name']==115)|\n",
    "        (data['Name']==561)|\n",
    "        (data['Name']==122)|\n",
    "        (data['Name']==331)|\n",
    "        (data['Name']==578)|\n",
    "        (data['Name']==405)|\n",
    "        (data['Name']==714)|\n",
    "        (data['Name']==812)|\n",
    "        (data['Name']==742)|\n",
    "        (data['Name']==423)|\n",
    "        (data['Name']==756)|\n",
    "        (data['Name']==321)|\n",
    "        (data['Name']==746)|\n",
    "        (data['Name']==208)|\n",
    "        (data['Name']==738)|\n",
    "        (data['Name']==278)|\n",
    "        (data['Name']==285)|\n",
    "        (data['Name']==306)|\n",
    "        (data['Name']==327)|\n",
    "        (data['Name']==350)|\n",
    "        (data['Name']==357)|\n",
    "        (data['Name']==379)|\n",
    "        (data['Name']==399)|\n",
    "        (data['Name']==433)|\n",
    "        (data['Name']==448)|\n",
    "        (data['Name']==744)|\n",
    "        (data['Name']==750)|\n",
    "        (data['Name']==821)|\n",
    "        (data['Name']==1016)|\n",
    "        (data['Name']==910)|\n",
    "        (data['Name']==933)|\n",
    "        (data['Name']==596)|\n",
    "        (data['Name']==675)|\n",
    "        (data['Name']==490)|\n",
    "        (data['Name']==349)|\n",
    "        (data['Name']==1028)|\n",
    "        (data['Name']==554)|\n",
    "        (data['Name']==712)|\n",
    "        (data['Name']==504)|\n",
    "        (data['Name']==813)|\n",
    "        (data['Name']==1024)|\n",
    "        (data['Name']==976)|\n",
    "        (data['Name']==72)|\n",
    "        (data['Name']==426)|\n",
    "        (data['Name']==108)|\n",
    "        (data['Name']==255)|\n",
    "        (data['Name']==326)|\n",
    "        (data['Name']==339)|\n",
    "        (data['Name']==449)|\n",
    "        (data['Name']==469)|\n",
    "        (data['Name']==971)|\n",
    "        (data['Name']==1017)|\n",
    "        (data['Name']==946)|\n",
    "        (data['Name']==575)|\n",
    "        (data['Name']==131)|\n",
    "        (data['Name']==487)|\n",
    "        (data['Name']==438)|\n",
    "        (data['Name']==669)|\n",
    "        (data['Name']==700)|\n",
    "        (data['Name']==329)|\n",
    "        (data['Name']==775)|\n",
    "        (data['Name']==842)|\n",
    "        (data['Name']==522)|\n",
    "        (data['Name']==972)|\n",
    "        (data['Name']==1014)|\n",
    "        (data['Name']==883)|\n",
    "        (data['Name']==912)|\n",
    "        (data['Name']==233)|\n",
    "        (data['Name']==717)|\n",
    "        (data['Name']==186)|\n",
    "        (data['Name']==32)|\n",
    "        (data['Name']==629)|\n",
    "        (data['Name']==370)|\n",
    "        (data['Name']==489)|\n",
    "        (data['Name']==354)|\n",
    "        (data['Name']==727)|\n",
    "        (data['Name']==291)|\n",
    "        (data['Name']==1026)|\n",
    "        (data['Name']==464)|\n",
    "        (data['Name']==236)|\n",
    "        (data['Name']==369)|\n",
    "        (data['Name']==754)|\n",
    "        (data['Name']==789)|\n",
    "        (data['Name']==962)|\n",
    "        (data['Name']==785)|\n",
    "        (data['Name']==926)|\n",
    "        (data['Name']==553)|\n",
    "        (data['Name']==359)|\n",
    "        (data['Name']==206)|\n",
    "        (data['Name']==827)|\n",
    "        (data['Name']==909)|\n",
    "        (data['Name']==320)|\n",
    "        (data['Name']==560)|\n",
    "        (data['Name']==116)|\n",
    "        (data['Name']==709)|\n",
    "        (data['Name']==414)|\n",
    "        (data['Name']==644)|\n",
    "        (data['Name']==605)|\n",
    "        (data['Name']==907)|\n",
    "        (data['Name']==806)|\n",
    "        (data['Name']==591)|\n",
    "        (data['Name']==479)|\n",
    "        (data['Name']==608)|\n",
    "        (data['Name']==788)|\n",
    "        (data['Name']==50)|\n",
    "        (data['Name']==757)|\n",
    "        (data['Name']==93)|\n",
    "        (data['Name']==109)|\n",
    "        (data['Name']==377)|\n",
    "        (data['Name']==127)|\n",
    "        (data['Name']==366)|\n",
    "        (data['Name']==861)|\n",
    "        (data['Name']==681)|\n",
    "        (data['Name']==801)|\n",
    "        (data['Name']==851)|\n",
    "        (data['Name']==930)|\n",
    "        (data['Name']==803)|\n",
    "        (data['Name']==76)|\n",
    "        (data['Name']==870)|\n",
    "        (data['Name']==63)|\n",
    "        (data['Name']==33)|\n",
    "        (data['Name']==777)|\n",
    "        (data['Name']==986)|\n",
    "        (data['Name']==1023)|\n",
    "        (data['Name']==284)|\n",
    "        (data['Name']==718)|\n",
    "        (data['Name']==724)|\n",
    "        (data['Name']==295)|\n",
    "        (data['Name']==783)|\n",
    "        (data['Name']==905)|\n",
    "        (data['Name']==476)|\n",
    "        (data['Name']==772)|\n",
    "        (data['Name']==809)|\n",
    "        (data['Name']==835)|\n",
    "        (data['Name']==164)|\n",
    "        (data['Name']==761)|\n",
    "        (data['Name']==825)|\n",
    "        (data['Name']==235)|\n",
    "        (data['Name']==488)|\n",
    "        (data['Name']==82)|\n",
    "        (data['Name']==473)|\n",
    "        (data['Name']==392)|\n",
    "        (data['Name']==896)|\n",
    "        (data['Name']==281)|\n",
    "        (data['Name']==299)|\n",
    "        (data['Name']==7)|\n",
    "        (data['Name']==607)|\n",
    "        (data['Name']==658)|\n",
    "        (data['Name']==382)|\n",
    "        (data['Name']==314)|\n",
    "        (data['Name']==170)|\n",
    "        (data['Name']==794)|\n",
    "        (data['Name']==720)|\n",
    "        (data['Name']==1011)|\n",
    "        (data['Name']==459)|\n",
    "        (data['Name']==990)|\n",
    "        (data['Name']==879)|\n",
    "        (data['Name']==732)|\n",
    "        (data['Name']==78)|\n",
    "        (data['Name']==787)|\n",
    "        (data['Name']==86)|\n",
    "        (data['Name']==916)|\n",
    "        (data['Name']==176)|\n",
    "        (data['Name']==297)|\n",
    "        (data['Name']==766)|\n",
    "        (data['Name']==531)|\n",
    "        (data['Name']==796)|\n",
    "        (data['Name']==1027)|\n",
    "        (data['Name']==982)|\n",
    "        (data['Name']==866)|\n",
    "        (data['Name']==336)|\n",
    "        (data['Name']==165)|\n",
    "        (data['Name']==740)|\n",
    "        (data['Name']==239)|\n",
    "        (data['Name']==646)|\n",
    "        (data['Name']==85)|\n",
    "        (data['Name']==755)|\n",
    "        (data['Name']==451)|\n",
    "        (data['Name']==220)|\n",
    "        (data['Name']==713)|\n",
    "        (data['Name']==418)|\n",
    "        (data['Name']==983)|\n",
    "        (data['Name']==380)|\n",
    "        (data['Name']==590)|\n",
    "        (data['Name']==432)|\n",
    "        (data['Name']==570)|\n",
    "        (data['Name']==463)|\n",
    "        (data['Name']==595)|\n",
    "        (data['Name']==529)|\n",
    "        (data['Name']==770)|\n",
    "        (data['Name']==395)|\n",
    "        (data['Name']==895)|\n",
    "        (data['Name']==214)|\n",
    "        (data['Name']==89)|\n",
    "        (data['Name']==793)|\n",
    "        (data['Name']==739)|\n",
    "        (data['Name']==741)|\n",
    "        (data['Name']==753)|\n",
    "        (data['Name']==460)|\n",
    "        (data['Name']==874)|\n",
    "        (data['Name']==484)|\n",
    "        (data['Name']==358)|\n",
    "        (data['Name']==29)|\n",
    "        (data['Name']==638)|\n",
    "        (data['Name']==346)|\n",
    "        (data['Name']==881)|\n",
    "        (data['Name']==376)|\n",
    "        (data['Name']==446)|\n",
    "        (data['Name']==38)|\n",
    "        (data['Name']==995)|\n",
    "        (data['Name']==949)|\n",
    "        (data['Name']==919)|\n",
    "        (data['Name']==167)|\n",
    "        (data['Name']==805)|\n",
    "        (data['Name']==980)|\n",
    "        (data['Name']==458)|\n",
    "        (data['Name']==580)|\n",
    "        (data['Name']==394)|\n",
    "        (data['Name']==443)|\n",
    "        (data['Name']==398)|\n",
    "        (data['Name']==886)|\n",
    "        (data['Name']==996)|\n",
    "        (data['Name']==465)|\n",
    "        (data['Name']==244)|\n",
    "        (data['Name']==703)|\n",
    "        (data['Name']==841)|\n",
    "        (data['Name']==138)|\n",
    "        (data['Name']==207)|\n",
    "        (data['Name']==19)|\n",
    "        (data['Name']==480)|\n",
    "        (data['Name']==411)|\n",
    "        (data['Name']==102)|\n",
    "        (data['Name']==663)|\n",
    "        (data['Name']==904)|\n",
    "        (data['Name']==767)|\n",
    "        (data['Name']==579)|\n",
    "        (data['Name']==974)|\n",
    "        (data['Name']==587)|\n",
    "        (data['Name']==344)|\n",
    "        (data['Name']==123)|\n",
    "        (data['Name']==441)|\n",
    "        (data['Name']==402)|\n",
    "        (data['Name']==814)|\n",
    "        (data['Name']==462)|\n",
    "        (data['Name']==921)|\n",
    "        (data['Name']==385)|\n",
    "        (data['Name']==778)|\n",
    "        (data['Name']==950)|\n",
    "        (data['Name']==655)|\n",
    "        (data['Name']==701)|\n",
    "        (data['Name']==795)|\n",
    "        (data['Name']==816)|\n",
    "        (data['Name']==181)|\n",
    "        (data['Name']==901)|\n",
    "        (data['Name']==936)|\n",
    "        (data['Name']==456)|\n",
    "        (data['Name']==885)|\n",
    "        (data['Name']==196)|\n",
    "        (data['Name']==294)|\n",
    "        (data['Name']==141)|\n",
    "        (data['Name']==408)|\n",
    "        (data['Name']==900)|\n",
    "        (data['Name']==630)|\n",
    "        (data['Name']==1000)|\n",
    "        (data['Name']==387)|\n",
    "        (data['Name']==152)|\n",
    "        (data['Name']==413)|\n",
    "        (data['Name']==915)|\n",
    "        (data['Name']==699)|\n",
    "        (data['Name']==890)|\n",
    "        (data['Name']==818)|\n",
    "        (data['Name']==128)|\n",
    "        (data['Name']==328)|\n",
    "        (data['Name']==270)|\n",
    "        (data['Name']==308)|\n",
    "        (data['Name']==749)|\n",
    "        (data['Name']==804)|\n",
    "        (data['Name']==631)|\n",
    "        (data['Name']==674)|\n",
    "        (data['Name']==889)|\n",
    "        (data['Name']==589)|\n",
    "        (data['Name']==144)|\n",
    "        (data['Name']==225)|\n",
    "        (data['Name']==158)|\n",
    "        (data['Name']==289)|\n",
    "        (data['Name']==654)|\n",
    "        (data['Name']==826)|\n",
    "        (data['Name']==272)|\n",
    "        (data['Name']==52)|\n",
    "        (data['Name']==637)|\n",
    "        (data['Name']==691)|\n",
    "        (data['Name']==689)|\n",
    "        (data['Name']==520)|\n",
    "        (data['Name']==657)|\n",
    "        (data['Name']==471)|\n",
    "        (data['Name']==166)|\n",
    "        (data['Name']==300)|\n",
    "        (data['Name']==765)|\n",
    "        (data['Name']==987)|\n",
    "        (data['Name']==639)|\n",
    "        (data['Name']==944)|\n",
    "        (data['Name']==994)|\n",
    "        (data['Name']==274)|\n",
    "        (data['Name']==254)|\n",
    "        (data['Name']==159)|\n",
    "        (data['Name']==240)|\n",
    "        (data['Name']==106)|\n",
    "        (data['Name']==530)|\n",
    "        (data['Name']==868)|\n",
    "        (data['Name']==429)|\n",
    "        (data['Name']==260)|\n",
    "        (data['Name']==417)|\n",
    "        (data['Name']==389)|\n",
    "        (data['Name']==690)|\n",
    "        (data['Name']==193)|\n",
    "        (data['Name']==685)|\n",
    "        (data['Name']==538)|\n",
    "        (data['Name']==406)|\n",
    "        (data['Name']==217)|\n",
    "        (data['Name']==648)|\n",
    "        (data['Name']==680)|\n",
    "        (data['Name']==798)|\n",
    "        (data['Name']==898)|\n",
    "        (data['Name']==677)|\n",
    "        (data['Name']==360)|\n",
    "        (data['Name']==163)|\n",
    "        (data['Name']==110)|\n",
    "        (data['Name']==386)|\n",
    "        (data['Name']==899)|\n",
    "        (data['Name']==991)|\n",
    "        (data['Name']==923)|\n",
    "        (data['Name']==692)|\n",
    "        (data['Name']==246)|\n",
    "        (data['Name']==647)|\n",
    "        (data['Name']==992)|\n",
    "        (data['Name']==172)|\n",
    "        (data['Name']==250)|\n",
    "        (data['Name']==894)|\n",
    "        (data['Name']==434)|\n",
    "        (data['Name']==182)|\n",
    "        (data['Name']==897)|\n",
    "        (data['Name']==993)|\n",
    "        (data['Name']==245)|\n",
    "        (data['Name']==751)|\n",
    "        (data['Name']==243)|\n",
    "        (data['Name']==997)|\n",
    "        (data['Name']==998)|\n",
    "        (data['Name']==988)|\n",
    "        (data['Name']==430)|\n",
    "        (data['Name']==435)|\n",
    "        (data['Name']==747)|\n",
    "        (data['Name']==305)|\n",
    "        (data['Name']==512)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**<font color='red'>Check if splitting was successful:</font>**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 334 entries, 0 to 333\n",
      "Data columns (total 1 columns):\n",
      "0    334 non-null int64\n",
      "dtypes: int64(1)\n",
      "memory usage: 2.7 KB\n",
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 334 entries, 2 to 1004\n",
      "Columns: 553 entries, Name to BCF\n",
      "dtypes: float64(375), int64(178)\n",
      "memory usage: 1.4 MB\n",
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 673 entries, 0 to 672\n",
      "Data columns (total 1 columns):\n",
      "0    673 non-null int64\n",
      "dtypes: int64(1)\n",
      "memory usage: 5.4 KB\n",
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 673 entries, 0 to 1006\n",
      "Columns: 553 entries, Name to BCF\n",
      "dtypes: float64(375), int64(178)\n",
      "memory usage: 2.8 MB\n"
     ]
    }
   ],
   "source": [
    "test_names.info()\n",
    "test.info()\n",
    "train_names.info()\n",
    "train.info()\n",
    "# Yes, everything correct!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# selected 3 features for Random Forest model as in the reference publication\n",
    "train = train[[\"TopoPSA\", \"WPATH\", \"XLogP3\", \"BCF\"]]\n",
    "test = test[[\"TopoPSA\", \"WPATH\", \"XLogP3\", \"BCF\"]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**<font color='red'>Model1: Imbalanced model towards nB-compounds:</font>** original - where negative labels are 4x more than positive labels (in the Training set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([532, 141])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# selecting features and labels for Random Forest model fitting\n",
    "X_train = train.drop(['BCF'], axis=1)\n",
    "y_train = train['BCF']\n",
    "X_test = test.drop(['BCF'], axis=1)\n",
    "y_test = test['BCF']\n",
    "\n",
    "import numpy as np\n",
    "np.bincount(y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    532\n",
      "1    141\n",
      "Name: BCF, dtype: int64\n",
      "0    266\n",
      "1     68\n",
      "Name: BCF, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# just to see how it looks like..\n",
    "print(y_train.value_counts())\n",
    "print(y_test.value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "rf = RandomForestClassifier(n_estimators=501, random_state=1234).fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix:\n",
      "[[242  24]\n",
      " [ 18  50]]\n"
     ]
    }
   ],
   "source": [
    "y_pred = rf.predict(X_test)\n",
    "confusion_matrix_result = confusion_matrix(y_test, y_pred)\n",
    "# label=[1,0], otherwise predicted and actual axes are swapped\n",
    "\n",
    "print(\"Confusion matrix:\\n%s\" % confusion_matrix_result)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "| ACTUAL\\PREDICTED          | non-Bioaccumulative (0) | Bioaccumulative (1) | TOTAL |\n",
    "|---------------------------|---------------------|-------------------------|-------|\n",
    "|    non-Bioaccumulative (0)    |      242             |        24               |   266   |\n",
    "|  Bioaccumulative (1)       |      18             |        50              |   68  |\n",
    "|        TOTAL              |      260             |        74              |   334  |"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.93      0.91      0.92       266\n",
      "           1       0.68      0.74      0.70        68\n",
      "\n",
      "    accuracy                           0.87       334\n",
      "   macro avg       0.80      0.82      0.81       334\n",
      "weighted avg       0.88      0.87      0.88       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred))\n",
    "\n",
    "# “specificity”: recall of the negative class\n",
    "# “sensitivity”: recall of the positive class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# result from original publication: MODEL_1 Imbalanced model towards nB-compounds\n",
    "from IPython.display import Image\n",
    "Image(filename='Imbalanced_nB.png')\n",
    "# our result is on the validation set\n",
    "## In the R implementation of Random Forests, there is a flag you can set to get the proximity matrix.\n",
    "## Proximity matrix is on training data with oob trees. (oob.prox=True in R)\n",
    "## There is nothing currently implemented for this in python.\n",
    "## https://stackoverflow.com/questions/18703136/proximity-matrix-in-sklearn-ensemble-randomforestclassifier\n",
    "## https://datascience.stackexchange.com/questions/17641/reproducing-randomforest-proximity-matrix-from-r-package-in-python"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='Acc_M1.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "specificity_M1: 0.00\n",
      "sensitivity_M1: 1.94\n",
      "accuracy_M1: 0.00\n"
     ]
    }
   ],
   "source": [
    "sp_M1=(0.91-0.91)\n",
    "se_M1=(0.74-0.7206)\n",
    "acc_M1=(0.87-0.87)\n",
    "sp=abs((0.91-0.91)*100)\n",
    "se=abs((0.74-0.7206)*100)\n",
    "acc=abs((0.87-0.87)*100)\n",
    "print(\"specificity_M1:\", \"%.2f\" % sp)\n",
    "print(\"sensitivity_M1:\", \"%.2f\" % se)\n",
    "print(\"accuracy_M1:\", \"%.2f\" % acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**<font color='red'>Out of bag prediction oob_score:</font>**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "about oob scores can read here: https://scikit-learn.org/stable/modules/ensemble.html\n",
    "but in principle, not clear how to reproduce the confusion matrix on training data with Sklearn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = train[[\"TopoPSA\", \"WPATH\", \"XLogP3\", \"BCF\"]]\n",
    "test = test[[\"TopoPSA\", \"WPATH\", \"XLogP3\", \"BCF\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = train.drop(['BCF'], axis=1)\n",
    "y_train = train['BCF']\n",
    "X_test = test.drop(['BCF'], axis=1)\n",
    "y_test = test['BCF']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf = RandomForestClassifier(n_estimators=501, oob_score=True, random_state=1234).fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8632986627043091"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf.oob_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix:\n",
      "[[532   0]\n",
      " [  0 141]]\n"
     ]
    }
   ],
   "source": [
    "y_pred = rf.predict(X_train)\n",
    "confusion_matrix_result = confusion_matrix(y_train, y_pred)\n",
    "# label=[1,0], otherwise predicted and actual axes are swapped\n",
    "\n",
    "print(\"Confusion matrix:\\n%s\" % confusion_matrix_result)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**<font color='red'>Model2: Balanced model:</font>** randomly took 141 out of 532 negative labels from the Train set to equalize to positive labels (141)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    532\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# BEFORE\n",
    "train.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    141\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Used undersampling and created a balanced training dataset\n",
    "# Selected all positive cases and randomly 141 negative cases\n",
    "# Show that the training dataset is balanced\n",
    "\n",
    "positive_cases = train[train['BCF'] == 1]\n",
    "negative_sample = train[train['BCF'] == 0].sample(141, random_state=0)\n",
    "train_balanced = pd.concat([positive_cases,negative_sample])\n",
    "train_balanced.sort_index(inplace=True)\n",
    "\n",
    "# split data into X and y\n",
    "X_train_bal = train_balanced.drop(['BCF'], axis=1)\n",
    "y_train_bal = train_balanced['BCF']\n",
    "\n",
    "### AFTER\n",
    "train_balanced.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([141, 141])"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# AFTER\n",
    "import numpy as np\n",
    "np.bincount(y_train_bal)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf = RandomForestClassifier(n_estimators=501, random_state=1234).fit(X_train_bal, y_train_bal)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix:\n",
      "[[221  45]\n",
      " [  6  62]]\n"
     ]
    }
   ],
   "source": [
    "y_pred = rf.predict(X_test)\n",
    "confusion_matrix_result = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "print(\"Confusion matrix:\\n%s\" % confusion_matrix_result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.97      0.83      0.90       266\n",
      "           1       0.58      0.91      0.71        68\n",
      "\n",
      "    accuracy                           0.85       334\n",
      "   macro avg       0.78      0.87      0.80       334\n",
      "weighted avg       0.89      0.85      0.86       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred))\n",
    "\n",
    "# “specificity”: recall of the negative class\n",
    "# “sensitivity”: recall of the positive class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# result from original publication: MODEL_2 Balanced model\n",
    "from IPython.display import Image\n",
    "Image(filename='Balanced.png')\n",
    "# our result is on the validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='Acc_M2.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "specificity_M2: 0.00\n",
      "sensitivity_M2: 1.29\n",
      "accuracy_M2: 1.00\n"
     ]
    }
   ],
   "source": [
    "sp_M2=(0.83-0.83)\n",
    "se_M2=(0.91-0.8971)\n",
    "acc_M2=(0.85-0.84)\n",
    "sp=abs((0.83-0.83)*100)\n",
    "se=abs((0.91-0.8971)*100)\n",
    "acc=abs((0.85-0.84)*100)\n",
    "print(\"specificity_M2:\", \"%.2f\" % sp)\n",
    "print(\"sensitivity_M2:\", \"%.2f\" % se)\n",
    "print(\"accuracy_M2:\", \"%.2f\" % acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**<font color='red'>Model2_2: Balanced:</font>** Used SMOTE technique (Synthetic Minority Oversampling Technique) that created equal amount of training instances (532) by over-sampling minority class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([532, 532])"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import imblearn\n",
    "from imblearn.over_sampling import SMOTE\n",
    "smote = SMOTE(random_state=0)\n",
    "X_train_resampled, y_train_resampled = smote.fit_sample(X_train, y_train)\n",
    "import numpy as np\n",
    "np.bincount(y_train_resampled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf = RandomForestClassifier(n_estimators=501, random_state=1234).fit(X_train_resampled, y_train_resampled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix:\n",
      "[[236  30]\n",
      " [ 11  57]]\n"
     ]
    }
   ],
   "source": [
    "y_pred = rf.predict(X_test)\n",
    "confusion_matrix_result = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "print(\"Confusion matrix:\\n%s\" % confusion_matrix_result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.89      0.92       266\n",
      "           1       0.66      0.84      0.74        68\n",
      "\n",
      "    accuracy                           0.88       334\n",
      "   macro avg       0.81      0.86      0.83       334\n",
      "weighted avg       0.89      0.88      0.88       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred))\n",
    "# “specificity”: recall of the negative class\n",
    "# “sensitivity”: recall of the positive class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "sp_M2_2=(0.89-0.83)\n",
    "se_M2_2=(0.84-0.8971)\n",
    "acc_M2_2=(0.88-0.84)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**<font color='red'>Model3: Imbalanced model towards B-compounds:</font>** randomly took 4:1 (positives:negatives) instances from Train set, e.g. looks like undersampling still"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    532\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check what is the distribution of our labels\n",
    "train.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    24\n",
       "1    94\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Selected randomly two-thirds of random B compounds (94 compounds)\n",
    "# and randomly a quarter from all B-compounds (24 compounds)\n",
    "# Show that the training dataset is balanced\n",
    "\n",
    "positive_cases = train[train['BCF'] == 1].sample(94, random_state=0)\n",
    "negative_sample = train[train['BCF'] == 0].sample(24, random_state=0)\n",
    "train_imbalanced = pd.concat([positive_cases,negative_sample])\n",
    "train_imbalanced.sort_index(inplace=True)\n",
    "\n",
    "# split data into X and y\n",
    "X_train_imbal = train_imbalanced.drop(['BCF'], axis=1)\n",
    "y_train_imbal = train_imbalanced['BCF']\n",
    "\n",
    "### TODO show that your training dataset is balanced\n",
    "train_imbalanced.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf = RandomForestClassifier(n_estimators=501, random_state=1234).fit(X_train_imbal, y_train_imbal)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix:\n",
      "[[159 107]\n",
      " [  0  68]]\n"
     ]
    }
   ],
   "source": [
    "y_pred = rf.predict(X_test)\n",
    "confusion_matrix_result = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "print(\"Confusion matrix:\\n%s\" % confusion_matrix_result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      0.60      0.75       266\n",
      "           1       0.39      1.00      0.56        68\n",
      "\n",
      "    accuracy                           0.68       334\n",
      "   macro avg       0.69      0.80      0.65       334\n",
      "weighted avg       0.88      0.68      0.71       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, y_pred))\n",
    "# “specificity”: recall of the negative class\n",
    "# “sensitivity”: recall of the positive class\n",
    "## Performance estimate F-measure would be more adequate here (Lecture 7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# result from original publication: MODEL_3 Imbalanced model towards B-compounds\n",
    "from IPython.display import Image\n",
    "Image(filename='Imbalanced_B.png')\n",
    "# our result is on the validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(filename='Acc_M3.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "specificity_M3: 6.92\n",
      "sensitivity_M3: 0.00\n",
      "accuracy_M3: 5.70\n"
     ]
    }
   ],
   "source": [
    "sp_M3=(0.60-0.6692)\n",
    "se_M3=(1.00-1.0000)\n",
    "acc_M3=(0.68-0.737)\n",
    "sp=abs((0.60-0.6692)*100)\n",
    "se=abs((1.00-1.0000)*100)\n",
    "acc=abs((0.68-0.737)*100)\n",
    "print(\"specificity_M3:\", \"%.2f\" % sp)\n",
    "print(\"sensitivity_M3:\", \"%.2f\" % se)\n",
    "print(\"accuracy_M3:\", \"%.2f\" % acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "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>Model</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>Error_sp_(decimal)</th>\n",
       "      <th>Sensitivity</th>\n",
       "      <th>Error_se_(decimal)</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>Error_acc_(decimal)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>M1: imbalanced_nB</td>\n",
       "      <td>0.91</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.74</td>\n",
       "      <td>0.0194</td>\n",
       "      <td>0.87</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>M2: balanced</td>\n",
       "      <td>0.83</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.91</td>\n",
       "      <td>0.0129</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>M2_2: balanced_SMOTE</td>\n",
       "      <td>0.89</td>\n",
       "      <td>0.0600</td>\n",
       "      <td>0.84</td>\n",
       "      <td>-0.0571</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.040</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>M3: imbalanced_B</td>\n",
       "      <td>0.60</td>\n",
       "      <td>-0.0692</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.68</td>\n",
       "      <td>-0.057</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  Model  Specificity  Error_sp_(decimal)  Sensitivity  \\\n",
       "0     M1: imbalanced_nB         0.91              0.0000         0.74   \n",
       "1          M2: balanced         0.83              0.0000         0.91   \n",
       "2  M2_2: balanced_SMOTE         0.89              0.0600         0.84   \n",
       "3      M3: imbalanced_B         0.60             -0.0692         1.00   \n",
       "\n",
       "   Error_se_(decimal)  Accuracy  Error_acc_(decimal)  \n",
       "0              0.0194      0.87                0.000  \n",
       "1              0.0129      0.85                0.010  \n",
       "2             -0.0571      0.88                0.040  \n",
       "3              0.0000      0.68               -0.057  "
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results = pd.DataFrame({ 'Model' : pd.Categorical([\"M1: imbalanced_nB\",\"M2: balanced\",\"M2_2: balanced_SMOTE\",\"M3: imbalanced_B\"]),\n",
    "                    'Specificity' : [0.91,0.83,0.89,0.6],\n",
    "                    'Error_sp_(decimal)' : [sp_M1,sp_M2,sp_M2_2,sp_M3],\n",
    "                    'Sensitivity' : [0.74,0.91,0.84,1.0],\n",
    "                    'Error_se_(decimal)' : [se_M1,se_M2,se_M2_2,se_M3],\n",
    "                    'Accuracy' : [0.87,0.85,0.88,0.68],\n",
    "                    'Error_acc_(decimal)' : [acc_M1, acc_M2, acc_M2_2, acc_M3]  })\n",
    "results\n",
    "## it is more important to find a model with higher sensitivity and \n",
    "## reasonable specificity for the classification of BCF!"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}