.. _solution2016creditclementrst:

===========================================================
2016 - Une solution à la compétition de machine learning 2A
===========================================================


.. only:: html

    **Links:** :download:`notebook <solution_2016_credit_clement.ipynb>`, :downloadlink:`html <solution_2016_credit_clement2html.html>`, :download:`python <solution_2016_credit_clement.py>`, :downloadlink:`slides <solution_2016_credit_clement.slides.html>`, :githublink:`GitHub|_doc/notebooks/notebook_eleves/2016-2017/solution_2016_credit_clement.ipynb|*`


Ce notebook a été proposé par un étudiant pour la compétition organisée
pour ce cours : `classification
binaire <http://www.xavierdupre.fr/app/ensae_teaching_cs/helpsphinx/all_notebooks.html#annee-2016>`__.

.. code:: ipython3

    from pyensae.datasource import download_data
    download_data("ensae_competition_2016.zip",
                  url="https://github.com/sdpython/ensae_teaching_cs/raw/master/_doc/competitions/2016_ENSAE_2A/")




.. parsed-literal::
    ['ensae_competition_test_X.txt', 'ensae_competition_train.txt']



.. code:: ipython3

    # packages
    import pandas as pd
    import numpy as np
    from sklearn import svm, linear_model, datasets, metrics
    import seaborn as sns
    import matplotlib.pyplot as plt
    %matplotlib inline
    from statsmodels.nonparametric.kde import KDEUnivariate
    from statsmodels.nonparametric import smoothers_lowess

.. code:: ipython3

    # dataframe
    # df = pd.read_excel("default_of_credit_card_clients.xls", header=[0, 1], encoding="utf8", index_col=0, engine='openpyxl')
    df = pd.read_csv("ensae_competition_train.txt", header=[0, 1], encoding="utf8", index_col=0, sep="\t")
    df.head(10)






.. raw:: html

    <div>
    <style>
        .dataframe thead tr:only-child th {
            text-align: right;
        }

        .dataframe thead th {
            text-align: left;
        }

        .dataframe tbody tr th {
            vertical-align: top;
        }
    </style>
    <table border="1" class="dataframe">
      <thead>
        <tr>
          <th></th>
          <th>X1</th>
          <th>X2</th>
          <th>X3</th>
          <th>X4</th>
          <th>X5</th>
          <th>X6</th>
          <th>X7</th>
          <th>X8</th>
          <th>X9</th>
          <th>X10</th>
          <th>...</th>
          <th>X15</th>
          <th>X16</th>
          <th>X17</th>
          <th>X18</th>
          <th>X19</th>
          <th>X20</th>
          <th>X21</th>
          <th>X22</th>
          <th>X23</th>
          <th>Y</th>
        </tr>
        <tr>
          <th>ID</th>
          <th>LIMIT_BAL</th>
          <th>SEX</th>
          <th>EDUCATION</th>
          <th>MARRIAGE</th>
          <th>AGE</th>
          <th>PAY_0</th>
          <th>PAY_2</th>
          <th>PAY_3</th>
          <th>PAY_4</th>
          <th>PAY_5</th>
          <th>...</th>
          <th>BILL_AMT4</th>
          <th>BILL_AMT5</th>
          <th>BILL_AMT6</th>
          <th>PAY_AMT1</th>
          <th>PAY_AMT2</th>
          <th>PAY_AMT3</th>
          <th>PAY_AMT4</th>
          <th>PAY_AMT5</th>
          <th>PAY_AMT6</th>
          <th>default payment next month</th>
        </tr>
      </thead>
      <tbody>
        <tr>
          <th>0</th>
          <td>180000</td>
          <td>1</td>
          <td>2</td>
          <td>1</td>
          <td>47</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>99694</td>
          <td>65977</td>
          <td>67415</td>
          <td>3700</td>
          <td>3700</td>
          <td>4100</td>
          <td>2360</td>
          <td>2500</td>
          <td>2618</td>
          <td>0</td>
        </tr>
        <tr>
          <th>1</th>
          <td>110000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>35</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>4869</td>
          <td>4966</td>
          <td>5070</td>
          <td>1053</td>
          <td>1073</td>
          <td>1081</td>
          <td>178</td>
          <td>184</td>
          <td>185</td>
          <td>1</td>
        </tr>
        <tr>
          <th>2</th>
          <td>70000</td>
          <td>2</td>
          <td>2</td>
          <td>2</td>
          <td>22</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>69927</td>
          <td>50579</td>
          <td>49483</td>
          <td>2501</td>
          <td>3001</td>
          <td>2608</td>
          <td>1777</td>
          <td>1792</td>
          <td>1793</td>
          <td>1</td>
        </tr>
        <tr>
          <th>3</th>
          <td>200000</td>
          <td>2</td>
          <td>1</td>
          <td>2</td>
          <td>27</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>...</td>
          <td>1665</td>
          <td>3370</td>
          <td>-36</td>
          <td>5610</td>
          <td>15616</td>
          <td>1673</td>
          <td>3385</td>
          <td>0</td>
          <td>95456</td>
          <td>0</td>
        </tr>
        <tr>
          <th>4</th>
          <td>370000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>39</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>48216</td>
          <td>47675</td>
          <td>48074</td>
          <td>2157</td>
          <td>2000</td>
          <td>1668</td>
          <td>2000</td>
          <td>3000</td>
          <td>1000</td>
          <td>0</td>
        </tr>
        <tr>
          <th>5</th>
          <td>260000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>29</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>-2</td>
          <td>-2</td>
          <td>...</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>3090</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>141516</td>
          <td>0</td>
        </tr>
        <tr>
          <th>6</th>
          <td>90000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>43</td>
          <td>-1</td>
          <td>-1</td>
          <td>2</td>
          <td>-1</td>
          <td>-1</td>
          <td>...</td>
          <td>7660</td>
          <td>21175</td>
          <td>4009</td>
          <td>4367</td>
          <td>9</td>
          <td>7660</td>
          <td>21175</td>
          <td>4009</td>
          <td>7452</td>
          <td>0</td>
        </tr>
        <tr>
          <th>7</th>
          <td>220000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>43</td>
          <td>-1</td>
          <td>3</td>
          <td>2</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>1090</td>
          <td>1090</td>
          <td>0</td>
          <td>167</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>1</td>
        </tr>
        <tr>
          <th>8</th>
          <td>50000</td>
          <td>1</td>
          <td>2</td>
          <td>1</td>
          <td>35</td>
          <td>1</td>
          <td>2</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>21260</td>
          <td>70</td>
          <td>29575</td>
          <td>0</td>
          <td>2052</td>
          <td>1800</td>
          <td>0</td>
          <td>29935</td>
          <td>1200</td>
          <td>1</td>
        </tr>
        <tr>
          <th>9</th>
          <td>50000</td>
          <td>2</td>
          <td>3</td>
          <td>2</td>
          <td>40</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>8292</td>
          <td>8465</td>
          <td>8650</td>
          <td>1271</td>
          <td>1130</td>
          <td>1000</td>
          <td>307</td>
          <td>325</td>
          <td>436</td>
          <td>0</td>
        </tr>
      </tbody>
    </table>
    <p>10 rows × 24 columns</p>
    </div>



.. code:: ipython3

    df.columns




.. parsed-literal::
    MultiIndex(levels=[['X1', 'X10', 'X11', 'X12', 'X13', 'X14', 'X15', 'X16', 'X17', 'X18', 'X19', 'X2', 'X20', 'X21', 'X22', 'X23', 'X3', 'X4', 'X5', 'X6', 'X7', 'X8', 'X9', 'Y'], ['AGE', 'BILL_AMT1', 'BILL_AMT2', 'BILL_AMT3', 'BILL_AMT4', 'BILL_AMT5', 'BILL_AMT6', 'EDUCATION', 'LIMIT_BAL', 'MARRIAGE', 'PAY_0', 'PAY_2', 'PAY_3', 'PAY_4', 'PAY_5', 'PAY_6', 'PAY_AMT1', 'PAY_AMT2', 'PAY_AMT3', 'PAY_AMT4', 'PAY_AMT5', 'PAY_AMT6', 'SEX', 'default payment next month']],
               labels=[[0, 11, 16, 17, 18, 19, 20, 21, 22, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 13, 14, 15, 23], [8, 22, 7, 9, 0, 10, 11, 12, 13, 14, 15, 1, 2, 3, 4, 5, 6, 16, 17, 18, 19, 20, 21, 23]],
               names=[None, 'ID'])



.. code:: ipython3

    # Retrait 2ème ligne header
    
    df1 = df.copy()
    df1.columns = df1.columns.droplevel(-1)

.. code:: ipython3

    df1.columns




.. parsed-literal::
    Index(['X1', 'X2', 'X3', 'X4', 'X5', 'X6', 'X7', 'X8', 'X9', 'X10', 'X11',
           'X12', 'X13', 'X14', 'X15', 'X16', 'X17', 'X18', 'X19', 'X20', 'X21',
           'X22', 'X23', 'Y'],
          dtype='object')



.. code:: ipython3

    # statistiques descriptives
    
    # paramètres des graphes
    fig = plt.figure(figsize=(12, 6)) 
    alpha=alpha_scatterplot = 0.2 
    alpha_bar_chart = 0.55
    
    '''graphs - the history of past payment'''
    
    # September 2005
    plt.subplot2grid((3,6),(0,0))
    plt.scatter(df1.Y, df1.X6, alpha=alpha_scatterplot)
    # axe x
    plt.xlabel("Default")
    # axe y
    plt.ylabel("Payment delay")
    # grid - titre                        
    plt.grid(b=True, which='major', axis='y')  
    plt.title("September 2005")
    
    # August 2005
    plt.subplot2grid((3,6),(0,1))
    plt.scatter(df1.Y, df1.X7, alpha=alpha_scatterplot)
    # axe x
    plt.xlabel("Default")
    # axe y
    plt.ylabel("Payment delay")
    # grid - titre                         
    plt.grid(b=True, which='major', axis='y')  
    plt.title("August 2005")
    
    # July 2005
    plt.subplot2grid((3,6),(0,2))
    plt.scatter(df1.Y, df1.X8, alpha=alpha_scatterplot)
    # axe x
    plt.xlabel("Default")
    # axe y
    plt.ylabel("Payment delay")
    # grid - titre                         
    plt.grid(b=True, which='major', axis='y')  
    plt.title("July 2005")
    
    # May 2005
    plt.subplot2grid((3,6),(0,3))
    plt.scatter(df1.Y, df1.X9, alpha=alpha_scatterplot)
    # axe x
    plt.xlabel("Default")
    # axe y
    plt.ylabel("Payment delay")
    # grid - titre                         
    plt.grid(b=True, which='major', axis='y')  
    plt.title("May 2005")
    
    # April 2005
    plt.subplot2grid((3,6),(0,4))
    plt.scatter(df1.Y, df1.X10, alpha=alpha_scatterplot)
    # axe x
    plt.xlabel("Default")
    # axe y
    plt.ylabel("Payment delay")
    # grid - titre             
    plt.grid(b=True, which='major', axis='y')  
    plt.title("April 2005")
    
    # March 2005
    plt.subplot2grid((3,6),(0,5))
    plt.scatter(df1.Y, df1.X11, alpha=alpha_scatterplot)
    # axe x
    plt.xlabel("Default")
    # axe y
    plt.ylabel("Payment delay")
    # grid - titre                          
    plt.grid(b=True, which='major', axis='y')  
    plt.title("March 2005")




.. parsed-literal::
    <matplotlib.text.Text at 0x2a5062587f0>




.. image:: solution_2016_credit_clement_7_1.png


.. code:: ipython3

    fig = plt.figure(figsize=(12, 6)) 
    alpha=alpha_scatterplot = 0.2 
    alpha_bar_chart = 0.55
    
    '''Graphs - bill statement'''
    
    # personnes pas en défaut de paiement
    ax1 = plt.subplot2grid((3,6),(1,0), colspan=3)
    # kernel density
    df1.X12[df1.Y == 0].plot(kind='kde')    
    df1.X13[df1.Y == 0].plot(kind='kde')
    df1.X14[df1.Y == 0].plot(kind='kde')
    df1.X15[df1.Y == 0].plot(kind='kde')
    df1.X16[df1.Y == 0].plot(kind='kde')
    df1.X17[df1.Y == 0].plot(kind='kde')
    # axes
    plt.xlabel("Bill statement")    
    plt.title("People distribution, no default")
    # limites
    ax1.set_xlim(0, 200000)
    # légende
    plt.legend(('September','August','July','May','April','March'),loc='best')
    
    # personnes en défaut de paiement
    ax2 = plt.subplot2grid((3,6),(1,3), colspan=3)
    # kernel density
    df1.X12[df1.Y == 1].plot(kind='kde')    
    df1.X13[df1.Y == 1].plot(kind='kde')
    df1.X14[df1.Y == 1].plot(kind='kde')
    df1.X15[df1.Y == 1].plot(kind='kde')
    df1.X16[df1.Y == 1].plot(kind='kde')
    df1.X17[df1.Y == 1].plot(kind='kde')
    # axes
    plt.xlabel("Bill statement")    
    plt.title("People distribution, default")
    # limites
    ax2.set_xlim(0, 200000)
    # légende
    plt.legend(('September','August','July','May','April','March'),loc='best')
    
    
    '''Graphs - amount of bill payed'''
    
    # personnes pas en défaut de paiement
    ax1 = plt.subplot2grid((3,6),(2,0), colspan=3)
    # kernel density
    df1.X18[df1.Y == 0].plot(kind='kde')    
    df1.X19[df1.Y == 0].plot(kind='kde')
    df1.X20[df1.Y == 0].plot(kind='kde')
    df1.X21[df1.Y == 0].plot(kind='kde')
    df1.X22[df1.Y == 0].plot(kind='kde')
    df1.X23[df1.Y == 0].plot(kind='kde')
    # axes
    plt.xlabel("Amount of bill payed")    
    plt.title("People distribution, no default")
    # limites
    ax1.set_xlim(0, 25000)
    # légende
    plt.legend(('September','August','July','May','April','March'),loc='best')
    
    # personnes en défaut de paiement
    ax2 = plt.subplot2grid((3,6),(2,3), colspan=3)
    # kernel density
    df1.X18[df1.Y == 1].plot(kind='kde')    
    df1.X19[df1.Y == 1].plot(kind='kde')
    df1.X20[df1.Y == 1].plot(kind='kde')
    df1.X21[df1.Y == 1].plot(kind='kde')
    df1.X22[df1.Y == 1].plot(kind='kde')
    df1.X23[df1.Y == 1].plot(kind='kde')
    # axes
    plt.xlabel("Amount of bill payed")    
    plt.title("People distribution, default")
    # limites
    ax2.set_xlim(0, 25000)
    # légende
    plt.legend(('September','August','July','May','April','March'),loc='best')




.. parsed-literal::
    <matplotlib.legend.Legend at 0x2a5065c50b8>




.. image:: solution_2016_credit_clement_8_1.png


.. code:: ipython3

    # Matrice des corrélations
    
    sns.set(context="paper", font="monospace")
    corrmat = df1.corr()
    
    # atplotlib figure
    f, ax = plt.subplots(figsize=(12, 9))
    
    # Draw the heatmap using seaborn
    sns.heatmap(corrmat, vmax=.8, square=True)




.. parsed-literal::
    <matplotlib.axes._subplots.AxesSubplot at 0x2a506461588>




.. image:: solution_2016_credit_clement_9_1.png


.. code:: ipython3

    # on modifie les colonnes (création de variables d'intérêt)
    
    df1['TotalDelay'] = df1.X11 + 2*df1.X10 + 4*df1.X9 + 8*df1.X8 + 16*df1.X7 + 32*df1.X6
    df1['TotalPayment'] = df1.X23 + 2*df1.X22 + 3*df1.X21 + 4*df1.X20 + 5*df1.X19 + 6*df1.X18
    df1['PartMay'] = -(df1.X22 - df1.X17)/(df1.X17 + 1)
    df1['PartJune'] = -(df1.X21 - df1.X16)/(df1.X16 + 1)
    df1['PartJuly'] = -(df1.X20 - df1.X15)/(df1.X15 + 1)
    df1['PartAugust'] = -(df1.X19 - df1.X14)/(df1.X14 + 1)
    df1['PartSeptember'] = -(df1.X18 - df1.X13)/(df1.X13 + 1)
    df1.head(20)






.. raw:: html

    <div>
    <style>
        .dataframe thead tr:only-child th {
            text-align: right;
        }

        .dataframe thead th {
            text-align: left;
        }

        .dataframe tbody tr th {
            vertical-align: top;
        }
    </style>
    <table border="1" class="dataframe">
      <thead>
        <tr style="text-align: right;">
          <th></th>
          <th>X1</th>
          <th>X2</th>
          <th>X3</th>
          <th>X4</th>
          <th>X5</th>
          <th>X6</th>
          <th>X7</th>
          <th>X8</th>
          <th>X9</th>
          <th>X10</th>
          <th>...</th>
          <th>X22</th>
          <th>X23</th>
          <th>Y</th>
          <th>TotalDelay</th>
          <th>TotalPayment</th>
          <th>PartMay</th>
          <th>PartJune</th>
          <th>PartJuly</th>
          <th>PartAugust</th>
          <th>PartSeptember</th>
        </tr>
      </thead>
      <tbody>
        <tr>
          <th>0</th>
          <td>180000</td>
          <td>1</td>
          <td>2</td>
          <td>1</td>
          <td>47</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>2500</td>
          <td>2618</td>
          <td>0</td>
          <td>0</td>
          <td>71798</td>
          <td>0.962902</td>
          <td>0.964215</td>
          <td>0.958865</td>
          <td>0.961978</td>
          <td>0.961112</td>
        </tr>
        <tr>
          <th>1</th>
          <td>110000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>35</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>184</td>
          <td>185</td>
          <td>1</td>
          <td>0</td>
          <td>17094</td>
          <td>0.963518</td>
          <td>0.963962</td>
          <td>0.777823</td>
          <td>0.721906</td>
          <td>0.850305</td>
        </tr>
        <tr>
          <th>2</th>
          <td>70000</td>
          <td>2</td>
          <td>2</td>
          <td>2</td>
          <td>22</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>1792</td>
          <td>1793</td>
          <td>1</td>
          <td>0</td>
          <td>51151</td>
          <td>0.963766</td>
          <td>0.964848</td>
          <td>0.962690</td>
          <td>0.956517</td>
          <td>0.962942</td>
        </tr>
        <tr>
          <th>3</th>
          <td>200000</td>
          <td>2</td>
          <td>1</td>
          <td>2</td>
          <td>27</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>...</td>
          <td>0</td>
          <td>95456</td>
          <td>0</td>
          <td>-126</td>
          <td>224043</td>
          <td>1.028571</td>
          <td>-0.004450</td>
          <td>-0.004802</td>
          <td>-0.004502</td>
          <td>-0.009899</td>
        </tr>
        <tr>
          <th>4</th>
          <td>370000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>39</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>3000</td>
          <td>1000</td>
          <td>0</td>
          <td>0</td>
          <td>42614</td>
          <td>0.937577</td>
          <td>0.958029</td>
          <td>0.965386</td>
          <td>0.959245</td>
          <td>0.968143</td>
        </tr>
        <tr>
          <th>5</th>
          <td>260000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>29</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>-2</td>
          <td>-2</td>
          <td>...</td>
          <td>0</td>
          <td>141516</td>
          <td>0</td>
          <td>-14</td>
          <td>160056</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.942813</td>
        </tr>
        <tr>
          <th>6</th>
          <td>90000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>43</td>
          <td>-1</td>
          <td>-1</td>
          <td>2</td>
          <td>-1</td>
          <td>-1</td>
          <td>...</td>
          <td>4009</td>
          <td>7452</td>
          <td>0</td>
          <td>-39</td>
          <td>135882</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.997711</td>
          <td>0.393333</td>
        </tr>
        <tr>
          <th>7</th>
          <td>220000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>43</td>
          <td>-1</td>
          <td>3</td>
          <td>2</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>0</td>
          <td>0</td>
          <td>1</td>
          <td>32</td>
          <td>1002</td>
          <td>0.000000</td>
          <td>0.999083</td>
          <td>0.999083</td>
          <td>0.999083</td>
          <td>0.866455</td>
        </tr>
        <tr>
          <th>8</th>
          <td>50000</td>
          <td>1</td>
          <td>2</td>
          <td>1</td>
          <td>35</td>
          <td>1</td>
          <td>2</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>29935</td>
          <td>1200</td>
          <td>1</td>
          <td>63</td>
          <td>78530</td>
          <td>-0.012172</td>
          <td>0.985915</td>
          <td>0.915291</td>
          <td>0.956269</td>
          <td>0.999979</td>
        </tr>
        <tr>
          <th>9</th>
          <td>50000</td>
          <td>2</td>
          <td>3</td>
          <td>2</td>
          <td>40</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>325</td>
          <td>436</td>
          <td>0</td>
          <td>0</td>
          <td>19283</td>
          <td>0.962316</td>
          <td>0.963619</td>
          <td>0.879296</td>
          <td>0.847636</td>
          <td>0.806216</td>
        </tr>
        <tr>
          <th>10</th>
          <td>130000</td>
          <td>1</td>
          <td>2</td>
          <td>1</td>
          <td>24</td>
          <td>0</td>
          <td>0</td>
          <td>2</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>0</td>
          <td>780</td>
          <td>0</td>
          <td>16</td>
          <td>38480</td>
          <td>0.997442</td>
          <td>0.999978</td>
          <td>0.959249</td>
          <td>0.999979</td>
          <td>0.899013</td>
        </tr>
        <tr>
          <th>11</th>
          <td>200000</td>
          <td>2</td>
          <td>1</td>
          <td>2</td>
          <td>25</td>
          <td>-1</td>
          <td>-1</td>
          <td>-1</td>
          <td>-1</td>
          <td>-1</td>
          <td>...</td>
          <td>4970</td>
          <td>8888</td>
          <td>0</td>
          <td>-63</td>
          <td>82748</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
        </tr>
        <tr>
          <th>12</th>
          <td>230000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>38</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>...</td>
          <td>2132</td>
          <td>2204</td>
          <td>0</td>
          <td>-126</td>
          <td>111453</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
        </tr>
        <tr>
          <th>13</th>
          <td>90000</td>
          <td>2</td>
          <td>1</td>
          <td>2</td>
          <td>29</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>-2</td>
          <td>...</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>-126</td>
          <td>0</td>
          <td>1.004184</td>
          <td>1.004184</td>
          <td>1.004184</td>
          <td>1.004184</td>
          <td>1.004184</td>
        </tr>
        <tr>
          <th>14</th>
          <td>230000</td>
          <td>1</td>
          <td>3</td>
          <td>2</td>
          <td>37</td>
          <td>-1</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>-1</td>
          <td>...</td>
          <td>5003</td>
          <td>3016</td>
          <td>0</td>
          <td>-34</td>
          <td>287273</td>
          <td>0.887873</td>
          <td>-0.000340</td>
          <td>0.769071</td>
          <td>0.760863</td>
          <td>0.952064</td>
        </tr>
        <tr>
          <th>15</th>
          <td>130000</td>
          <td>1</td>
          <td>2</td>
          <td>2</td>
          <td>33</td>
          <td>2</td>
          <td>2</td>
          <td>-1</td>
          <td>-1</td>
          <td>-2</td>
          <td>...</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>78</td>
          <td>3578</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.993003</td>
        </tr>
        <tr>
          <th>16</th>
          <td>90000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>35</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>4000</td>
          <td>0</td>
          <td>0</td>
          <td>2</td>
          <td>91108</td>
          <td>0.954684</td>
          <td>0.883510</td>
          <td>0.952754</td>
          <td>0.953646</td>
          <td>0.952596</td>
        </tr>
        <tr>
          <th>17</th>
          <td>10000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>37</td>
          <td>-1</td>
          <td>4</td>
          <td>3</td>
          <td>2</td>
          <td>2</td>
          <td>...</td>
          <td>0</td>
          <td>36</td>
          <td>0</td>
          <td>70</td>
          <td>3236</td>
          <td>0.999550</td>
          <td>0.848221</td>
          <td>0.800478</td>
          <td>0.999590</td>
          <td>0.999652</td>
        </tr>
        <tr>
          <th>18</th>
          <td>80000</td>
          <td>1</td>
          <td>3</td>
          <td>1</td>
          <td>36</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>3000</td>
          <td>6200</td>
          <td>0</td>
          <td>0</td>
          <td>73411</td>
          <td>0.961214</td>
          <td>0.926935</td>
          <td>0.962670</td>
          <td>0.963936</td>
          <td>0.962893</td>
        </tr>
        <tr>
          <th>19</th>
          <td>320000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>36</td>
          <td>-1</td>
          <td>2</td>
          <td>0</td>
          <td>0</td>
          <td>0</td>
          <td>...</td>
          <td>5000</td>
          <td>11906</td>
          <td>0</td>
          <td>0</td>
          <td>96906</td>
          <td>0.792507</td>
          <td>0.755381</td>
          <td>0.703680</td>
          <td>0.400943</td>
          <td>0.999851</td>
        </tr>
      </tbody>
    </table>
    <p>20 rows × 31 columns</p>
    </div>



.. code:: ipython3

    # Matrice des corrélations
    
    sns.set(context="paper", font="monospace")
    corrmat = df1.corr()
    
    # matplotlib figure
    f, ax = plt.subplots(figsize=(12, 9))
    
    # Draw the heatmap using seaborn
    sns.heatmap(corrmat, vmax=.8, square=True)




.. parsed-literal::
    <matplotlib.axes._subplots.AxesSubplot at 0x2a5067c70f0>




.. image:: solution_2016_credit_clement_11_1.png


.. code:: ipython3

    # drop some columns
    
    df1 = df1.drop(['X'+str(n) for n in range(7,12)] + ['X'+str(n) for n in range(13,24)], axis=1)
    df1.head(20)






.. raw:: html

    <div>
    <style>
        .dataframe thead tr:only-child th {
            text-align: right;
        }

        .dataframe thead th {
            text-align: left;
        }

        .dataframe tbody tr th {
            vertical-align: top;
        }
    </style>
    <table border="1" class="dataframe">
      <thead>
        <tr style="text-align: right;">
          <th></th>
          <th>X1</th>
          <th>X2</th>
          <th>X3</th>
          <th>X4</th>
          <th>X5</th>
          <th>X6</th>
          <th>X12</th>
          <th>Y</th>
          <th>TotalDelay</th>
          <th>TotalPayment</th>
          <th>PartMay</th>
          <th>PartJune</th>
          <th>PartJuly</th>
          <th>PartAugust</th>
          <th>PartSeptember</th>
        </tr>
      </thead>
      <tbody>
        <tr>
          <th>0</th>
          <td>180000</td>
          <td>1</td>
          <td>2</td>
          <td>1</td>
          <td>47</td>
          <td>0</td>
          <td>179253</td>
          <td>0</td>
          <td>0</td>
          <td>71798</td>
          <td>0.962902</td>
          <td>0.964215</td>
          <td>0.958865</td>
          <td>0.961978</td>
          <td>0.961112</td>
        </tr>
        <tr>
          <th>1</th>
          <td>110000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>35</td>
          <td>0</td>
          <td>6137</td>
          <td>1</td>
          <td>0</td>
          <td>17094</td>
          <td>0.963518</td>
          <td>0.963962</td>
          <td>0.777823</td>
          <td>0.721906</td>
          <td>0.850305</td>
        </tr>
        <tr>
          <th>2</th>
          <td>70000</td>
          <td>2</td>
          <td>2</td>
          <td>2</td>
          <td>22</td>
          <td>0</td>
          <td>66505</td>
          <td>1</td>
          <td>0</td>
          <td>51151</td>
          <td>0.963766</td>
          <td>0.964848</td>
          <td>0.962690</td>
          <td>0.956517</td>
          <td>0.962942</td>
        </tr>
        <tr>
          <th>3</th>
          <td>200000</td>
          <td>2</td>
          <td>1</td>
          <td>2</td>
          <td>27</td>
          <td>-2</td>
          <td>4941</td>
          <td>0</td>
          <td>-126</td>
          <td>224043</td>
          <td>1.028571</td>
          <td>-0.004450</td>
          <td>-0.004802</td>
          <td>-0.004502</td>
          <td>-0.009899</td>
        </tr>
        <tr>
          <th>4</th>
          <td>370000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>39</td>
          <td>0</td>
          <td>141552</td>
          <td>0</td>
          <td>0</td>
          <td>42614</td>
          <td>0.937577</td>
          <td>0.958029</td>
          <td>0.965386</td>
          <td>0.959245</td>
          <td>0.968143</td>
        </tr>
        <tr>
          <th>5</th>
          <td>260000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>29</td>
          <td>0</td>
          <td>71864</td>
          <td>0</td>
          <td>-14</td>
          <td>160056</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.942813</td>
        </tr>
        <tr>
          <th>6</th>
          <td>90000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>43</td>
          <td>-1</td>
          <td>16139</td>
          <td>0</td>
          <td>-39</td>
          <td>135882</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.997711</td>
          <td>0.393333</td>
        </tr>
        <tr>
          <th>7</th>
          <td>220000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>43</td>
          <td>-1</td>
          <td>1090</td>
          <td>1</td>
          <td>32</td>
          <td>1002</td>
          <td>0.000000</td>
          <td>0.999083</td>
          <td>0.999083</td>
          <td>0.999083</td>
          <td>0.866455</td>
        </tr>
        <tr>
          <th>8</th>
          <td>50000</td>
          <td>1</td>
          <td>2</td>
          <td>1</td>
          <td>35</td>
          <td>1</td>
          <td>48047</td>
          <td>1</td>
          <td>63</td>
          <td>78530</td>
          <td>-0.012172</td>
          <td>0.985915</td>
          <td>0.915291</td>
          <td>0.956269</td>
          <td>0.999979</td>
        </tr>
        <tr>
          <th>9</th>
          <td>50000</td>
          <td>2</td>
          <td>3</td>
          <td>2</td>
          <td>40</td>
          <td>0</td>
          <td>5538</td>
          <td>0</td>
          <td>0</td>
          <td>19283</td>
          <td>0.962316</td>
          <td>0.963619</td>
          <td>0.879296</td>
          <td>0.847636</td>
          <td>0.806216</td>
        </tr>
        <tr>
          <th>10</th>
          <td>130000</td>
          <td>1</td>
          <td>2</td>
          <td>1</td>
          <td>24</td>
          <td>0</td>
          <td>46113</td>
          <td>0</td>
          <td>16</td>
          <td>38480</td>
          <td>0.997442</td>
          <td>0.999978</td>
          <td>0.959249</td>
          <td>0.999979</td>
          <td>0.899013</td>
        </tr>
        <tr>
          <th>11</th>
          <td>200000</td>
          <td>2</td>
          <td>1</td>
          <td>2</td>
          <td>25</td>
          <td>-1</td>
          <td>8926</td>
          <td>0</td>
          <td>-63</td>
          <td>82748</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
        </tr>
        <tr>
          <th>12</th>
          <td>230000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>38</td>
          <td>-2</td>
          <td>12696</td>
          <td>0</td>
          <td>-126</td>
          <td>111453</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
        </tr>
        <tr>
          <th>13</th>
          <td>90000</td>
          <td>2</td>
          <td>1</td>
          <td>2</td>
          <td>29</td>
          <td>-2</td>
          <td>-240</td>
          <td>0</td>
          <td>-126</td>
          <td>0</td>
          <td>1.004184</td>
          <td>1.004184</td>
          <td>1.004184</td>
          <td>1.004184</td>
          <td>1.004184</td>
        </tr>
        <tr>
          <th>14</th>
          <td>230000</td>
          <td>1</td>
          <td>3</td>
          <td>2</td>
          <td>37</td>
          <td>-1</td>
          <td>36571</td>
          <td>0</td>
          <td>-34</td>
          <td>287273</td>
          <td>0.887873</td>
          <td>-0.000340</td>
          <td>0.769071</td>
          <td>0.760863</td>
          <td>0.952064</td>
        </tr>
        <tr>
          <th>15</th>
          <td>130000</td>
          <td>1</td>
          <td>2</td>
          <td>2</td>
          <td>33</td>
          <td>2</td>
          <td>2183</td>
          <td>0</td>
          <td>78</td>
          <td>3578</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.000000</td>
          <td>0.993003</td>
        </tr>
        <tr>
          <th>16</th>
          <td>90000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>35</td>
          <td>0</td>
          <td>72112</td>
          <td>0</td>
          <td>2</td>
          <td>91108</td>
          <td>0.954684</td>
          <td>0.883510</td>
          <td>0.952754</td>
          <td>0.953646</td>
          <td>0.952596</td>
        </tr>
        <tr>
          <th>17</th>
          <td>10000</td>
          <td>2</td>
          <td>2</td>
          <td>1</td>
          <td>37</td>
          <td>-1</td>
          <td>3305</td>
          <td>0</td>
          <td>70</td>
          <td>3236</td>
          <td>0.999550</td>
          <td>0.848221</td>
          <td>0.800478</td>
          <td>0.999590</td>
          <td>0.999652</td>
        </tr>
        <tr>
          <th>18</th>
          <td>80000</td>
          <td>1</td>
          <td>3</td>
          <td>1</td>
          <td>36</td>
          <td>0</td>
          <td>81066</td>
          <td>0</td>
          <td>0</td>
          <td>73411</td>
          <td>0.961214</td>
          <td>0.926935</td>
          <td>0.962670</td>
          <td>0.963936</td>
          <td>0.962893</td>
        </tr>
        <tr>
          <th>19</th>
          <td>320000</td>
          <td>2</td>
          <td>1</td>
          <td>1</td>
          <td>36</td>
          <td>-1</td>
          <td>7868</td>
          <td>0</td>
          <td>0</td>
          <td>96906</td>
          <td>0.792507</td>
          <td>0.755381</td>
          <td>0.703680</td>
          <td>0.400943</td>
          <td>0.999851</td>
        </tr>
      </tbody>
    </table>
    </div>



.. code:: ipython3

    from sklearn.decomposition import PCA
    from numpy import inf
    pca = PCA(n_components=2, svd_solver='randomized')
    dfpca = df1.values
    dfpca[dfpca == -inf] = 0
    y = dfpca[:, 7]
    proj = pca.fit_transform(dfpca[:, :7 + 8:])
    plt.scatter(proj[:, 0], proj[:, 1], c=y)
    plt.colorbar()




.. parsed-literal::
    <matplotlib.colorbar.Colorbar at 0x2a506a1a0b8>




.. image:: solution_2016_credit_clement_13_1.png


.. code:: ipython3

    # training/crossval set
    
    X = df1.values
    X[X==-inf] = 0
    print(df1.head())
    # training set
    X_train = X[:, :]
    Y_train = X[:, 7].ravel()
    X_train = np.delete(X_train, 7, axis=1)
    # expected result
    expected = X[20000:, 7].ravel()
    # cross-validation data set
    X_cross = X[20000:, :]
    X_cross = np.delete(X_cross, 7, axis=1)


.. parsed-literal::
           X1  X2  X3  X4  X5  X6     X12  Y  TotalDelay  TotalPayment   PartMay  \
    0  180000   1   2   1  47   0  179253  0           0         71798  0.962902   
    1  110000   2   2   1  35   0    6137  1           0         17094  0.963518   
    2   70000   2   2   2  22   0   66505  1           0         51151  0.963766   
    3  200000   2   1   2  27  -2    4941  0        -126        224043  1.028571   
    4  370000   2   1   1  39   0  141552  0           0         42614  0.937577   
       PartJune  PartJuly  PartAugust  PartSeptember  
    0  0.964215  0.958865    0.961978       0.961112  
    1  0.963962  0.777823    0.721906       0.850305  
    2  0.964848  0.962690    0.956517       0.962942  
    3 -0.004450 -0.004802   -0.004502      -0.009899  
    4  0.958029  0.965386    0.959245       0.968143  


.. code:: ipython3

    from sklearn.naive_bayes import GaussianNB
    
    # train the model
    GNB = GaussianNB()
    GNB.fit(X_train, Y_train)
    
    # use the model to predict the labels of the test data
    predicted = GNB.predict(X_cross)
    
    print(metrics.confusion_matrix(expected, predicted))


.. parsed-literal::
    [[ 209 1732]
     [  26  533]]


.. code:: ipython3

    from sklearn.ensemble import GradientBoostingClassifier
    
    GBR = GradientBoostingClassifier()
    GBR.fit(X_train,Y_train)
    predicted = GBR.predict(X_cross)
    print(metrics.confusion_matrix(expected, predicted))


.. parsed-literal::
    [[1848   93]
     [ 352  207]]


.. code:: ipython3

    from sklearn.neighbors import KNeighborsClassifier
    
    KNC = KNeighborsClassifier(5)
    KNC.fit(X_train, Y_train)
    predicted = KNC.predict(X_cross)
    print(metrics.confusion_matrix(expected, predicted))
    pred = KNC.predict_proba(X_train)


.. parsed-literal::
    [[1864   77]
     [ 368  191]]


.. code:: ipython3

    print(pred[:10])
    print(Y_train[:10])


.. parsed-literal::
    [[ 1.   0. ]
     [ 0.6  0.4]
     [ 0.6  0.4]
     [ 1.   0. ]
     [ 1.   0. ]
     [ 0.6  0.4]
     [ 1.   0. ]
     [ 0.4  0.6]
     [ 0.4  0.6]
     [ 0.8  0.2]]
    [ 0.  1.  1.  0.  0.  0.  0.  1.  1.  0.]


.. code:: ipython3

    # neural network
    
    from sklearn.neural_network import MLPClassifier
    from sklearn.metrics import r2_score
    from sklearn.model_selection import GridSearchCV
    
    # optimisation - choix du nombre de couches
    param_grid = [
      {'hidden_layer_sizes': [(nb,) for nb in range(20,50,10)]}, 
      {'alpha': [a/100 for a in range(0,40,20)]}
     ]
    
    neural2 = GridSearchCV(MLPClassifier(), param_grid, verbose=1)
    neural2.fit(X_train, Y_train)
    neural2.best_estimator_


.. parsed-literal::
    Fitting 3 folds for each of 5 candidates, totalling 15 fits


.. parsed-literal::

    [Parallel(n_jobs=1)]: Done  15 out of  15 | elapsed:   27.2s finished




.. parsed-literal::
    MLPClassifier(activation='relu', alpha=0.0001, batch_size='auto', beta_1=0.9,
           beta_2=0.999, early_stopping=False, epsilon=1e-08,
           hidden_layer_sizes=(20,), learning_rate='constant',
           learning_rate_init=0.001, max_iter=200, momentum=0.9,
           nesterovs_momentum=True, power_t=0.5, random_state=None,
           shuffle=True, solver='adam', tol=0.0001, validation_fraction=0.1,
           verbose=False, warm_start=False)



.. code:: ipython3

    neural = MLPClassifier(activation='relu', alpha=0.0001, batch_size='auto', beta_1=0.9,
           beta_2=0.999, early_stopping=False, epsilon=1e-08,
           hidden_layer_sizes=(170,), learning_rate='constant',
           learning_rate_init=0.001, max_iter=200, momentum=0.9,
           nesterovs_momentum=True, power_t=0.5, random_state=None,
           shuffle=True, solver='adam', tol=0.0001, validation_fraction=0.1,
           verbose=False, warm_start=False)
    neural.fit(X_train, Y_train)
    predicted = neural.predict(X_cross)
    print(metrics.confusion_matrix(expected, predicted))
    neural.predict_proba(X_cross[:10])


.. parsed-literal::
    [[1780  161]
     [ 378  181]]




.. parsed-literal::

    array([[  1.00000000e+000,   4.54194423e-294],
           [  1.00000000e+000,   1.14673239e-101],
           [  1.00000000e+000,   1.13397258e-051],
           [  1.00000000e+000,   1.89540529e-117],
           [  1.00000000e+000,   8.01811448e-032],
           [  1.00000000e+000,   1.14003085e-160],
           [  1.00000000e+000,   1.02562443e-115],
           [  1.00000000e+000,   9.41507727e-017],
           [  1.00000000e+000,   3.16744761e-026],
           [  6.95744980e-001,   3.04255020e-001]])



.. code:: ipython3

    if_you_have_time = False
    if if_you_have_time:
        from sklearn.gaussian_process import GaussianProcessClassifier
    
        GPC = GaussianProcessClassifier()
        GPC.fit(X_train, Y_train)
        predicted = GPC.predict(X_cross)
        print(metrics.confusion_matrix(expected, predicted))
        GPC.predict_proba(X_cross)

.. code:: ipython3

    from sklearn.ensemble import RandomForestClassifier
    
    RFC = RandomForestClassifier(5)
    RFC.fit(X_train, Y_train)
    predicted = RFC.predict(X_cross)
    print(metrics.confusion_matrix(expected, predicted))
    print(RFC.predict_proba(X_cross))
    print(expected)


.. parsed-literal::
    [[1928   13]
     [  65  494]]
    [[ 1.   0. ]
     [ 1.   0. ]
     [ 0.   1. ]
     ..., 
     [ 1.   0. ]
     [ 1.   0. ]
     [ 0.8  0.2]]
    [ 0.  0.  1. ...,  0.  0.  0.]


.. code:: ipython3

    if if_you_have_time:
        from sklearn.svm import SVC
    
        SVC = SVC(probability = True)
        SVC.fit(X_train, Y_train)
        predicted = SVC.predict(X_cross)
        print(metrics.confusion_matrix(expected, predicted))
        print(SVC.predict_proba(X_cross))
        print(expected)

.. code:: ipython3

    from sklearn.linear_model import LogisticRegression
    
    LR = LogisticRegression()
    LR.fit(X_train, Y_train)
    predicted = LR.predict(X_cross)
    print(metrics.confusion_matrix(expected, predicted))
    print(LR.predict_proba(X_train))
    print(Y_train)


.. parsed-literal::
    [[1861   80]
     [ 413  146]]
    [[ 0.85603222  0.14396778]
     [ 0.72020052  0.27979948]
     [ 0.69524708  0.30475292]
     ..., 
     [ 0.69884026  0.30115974]
     [ 0.85309221  0.14690779]
     [ 0.75099513  0.24900487]]
    [ 0.  1.  1. ...,  0.  0.  0.]


.. code:: ipython3

    #--------------#
    # modèle final #
    #--------------#
    
    # dataframe
    dfend = pd.read_csv("ensae_competition_test_X.txt", header=[0, 1], sep='\t', encoding="utf8", index_col=0)
    dfend.columns = dfend.columns.droplevel(-1)
    
    # modifications colonnes
    dfend['TotalDelay'] = dfend.X11 + 2*dfend.X10 + 4*dfend.X9 + 8*dfend.X8 + 16*dfend.X7 + 32*dfend.X6
    dfend['TotalPayment'] = dfend.X23 + 2*dfend.X22 + 3*dfend.X21 + 4*dfend.X20 + 5*dfend.X19 + 6*dfend.X18
    dfend['PartMay'] = -(dfend.X22 - dfend.X17)/(dfend.X17 + 1)
    dfend['PartJune'] = -(dfend.X21 - dfend.X16)/(dfend.X16 + 1)
    dfend['PartJuly'] = -(dfend.X20 - dfend.X15)/(dfend.X15 + 1)
    dfend['PartAugust'] = -(dfend.X19 - dfend.X14)/(dfend.X14 + 1)
    dfend['PartSeptember'] = -(dfend.X18 - dfend.X13)/(dfend.X13 + 1)
    dfend = dfend.drop(['X'+str(n) for n in range(7,12)] + ['X'+str(n) for n in range(13,24)], axis=1)
    
    # dataset as array
    X = dfend.values
    X[X==-inf] = 0

.. code:: ipython3

    # prédictions
    
    # réseau de neuronnes
    l = neural.predict(X)
    text_file = open('answerN.txt','w')
    for e in l:
        text_file.write(str(int(e)) + '\n')
    text_file.close()

.. code:: ipython3

    # random forest
    l = RFC.predict(X)
    text_file = open('answerRF.txt','w')
    for e in l:
        text_file.write(str(int(e)) + '\n')
    text_file.close()