EDA



In [ ]:

    
import pandas as pd
import pandas_profiling
import matplotlib
from datetime import datetime
import numpy as np
import matplotlib.pyplot as plt

matplotlib.style.use('ggplot')
%matplotlib inline

Chargement des données



In [213]:

    
df_bc = pd.read_csv('data/dataset_betclic.csv')



In [215]:

    
df_bc['BirthDate'] = pd.to_datetime(df_bc['BirthDate'], format='%Y-%m-%dT%H:%M:%S.%fZ')
df_bc['FirstDepositDate'] = pd.to_datetime(df_bc['FirstDepositDate'], format='%Y-%m-%dT%H:%M:%S.%fZ')
df_bc['BetDate'] = pd.to_datetime(df_bc['BetDate'], format='%Y%m%d')

Description de la donnée

Un outil parfait pour les EDA mais très verbeux :

pandas_profiling.ProfileReport(df_bc)

je prefere ne pas le lancer ici et explorer à la main

Country



In [7]:

    
df_bc.groupby('Country')['UserId'].count().sort_values(ascending=True).plot(kind='barh', figsize=(20, 5))









    Out[7]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f41be08fb38>

Gender



In [41]:

    
df_bc.groupby('Gender')['UserId'].count().sort_values(ascending=False).plot(kind='pie')









    Out[41]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f41aa4dbb70>

Birthdate



In [67]:

    
df_bc['BirthDate'].hist(bins=100, figsize=(20, 5))









    Out[67]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8da423c88>

Partner type



In [68]:

    
df_bc.groupby('PartnerType')['UserId'].count().sort_values(ascending=True).plot(kind='barh', figsize=(20, 5))









    Out[68]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8da2bc978>

First deposit date



In [75]:

    
df_bc['FirstDepositDate'].hist(bins=100, figsize=(20, 5))









    Out[75]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8da05f668>

Bet date



In [70]:

    
df_bc['BetDate'].hist(bins=100, figsize=(20, 5))









    Out[70]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8da1cb2b0>



In [80]:

    
df_bc.groupby('BetDate')['BetId'].count().plot(kind='hist', bins=100, figsize=(20, 5))









    Out[80]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8fb7aaa58>

Application



In [81]:

    
df_bc.groupby('Application')['UserId'].count().sort_values(ascending=True).plot(kind='barh', figsize=(20, 5))









    Out[81]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8fb6f19b0>

Product name



In [82]:

    
df_bc.groupby('ProductName')['UserId'].count().sort_values(ascending=True).plot(kind='barh', figsize=(20, 5))









    Out[82]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8fb662cf8>

Is live



In [42]:

    
df_bc.groupby('IsLive')['UserId'].count().sort_values(ascending=True).plot(kind='pie', figsize=(20, 5))









    Out[42]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f41aa2e4358>

Bet Amount



In [15]:

    
df_bc['BetAmount'].hist(bins=100, bottom=0.1, figsize=(20, 5))









    Out[15]:





<matplotlib.axes._subplots.AxesSubplot at 0x7efc408465f8>

il est impossible de voir quoi que ce soit ici.
Un changement de referentiel permet d'observer la distribution des bets amount



In [12]:

    
fig, ax = plt.subplots()
df_bc['BetAmount'].hist(ax=ax, bins=100, bottom=0.1, figsize=(20, 5))
ax.set_yscale('log')

Amount won (mode log)



In [90]:

    
fig, ax = plt.subplots()
df_bc['AmountWon'].hist(ax=ax, bins=100, bottom=0.1, figsize=(20, 5))
ax.set_yscale('log')

Odds



In [92]:

    
fig, ax = plt.subplots()
df_bc['Odds'].hist(ax=ax, bins=100, bottom=0.1, figsize=(20, 5))
ax.set_yscale('log')

Cash out



In [94]:

    
fig, ax = plt.subplots()
df_bc['Cashout'].hist(ax=ax, bins=100, bottom=0.1, figsize=(20, 5))
ax.set_yscale('log')

Question

quel est l'age moyen d'un client ?



In [107]:

    
(df_bc.drop_duplicates(subset='UserId')['BirthDate'] - df_bc.drop_duplicates(subset='UserId')['BirthDate'].min()).mean() + df_bc.drop_duplicates(subset='UserId')['BirthDate'].min()









    Out[107]:





Timestamp('1988-05-17 03:48:01.834728960')



In [220]:

    
now = pd.Timestamp(datetime.now())
(now - df_bc.drop_duplicates(subset='UserId')['BirthDate']).astype('<m8[Y]').mean()









    Out[220]:





30.242082271568986

Des femmes ?



In [21]:

    
(df_bc.drop_duplicates(subset='UserId')[df_bc['Gender']=='F']['BirthDate'] - df_bc.drop_duplicates(subset='UserId')[df_bc['Gender']=='F']['BirthDate'].min()).mean() + df_bc.drop_duplicates(subset='UserId')[df_bc['Gender']=='F']['BirthDate'].min()









    



/home/tom/source/smart_click/betclic/lib/python3.5/site-packages/ipykernel_launcher.py:1: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  """Entry point for launching an IPython kernel.






    Out[21]:





Timestamp('1987-05-12 16:17:10.351990528')

Des hommes ?



In [22]:

    
(df_bc.drop_duplicates(subset='UserId')[df_bc['Gender']=='M']['BirthDate'] - df_bc.drop_duplicates(subset='UserId')[df_bc['Gender']=='M']['BirthDate'].min()).mean() + df_bc.drop_duplicates(subset='UserId')[df_bc['Gender']=='M']['BirthDate'].min()









    



/home/tom/source/smart_click/betclic/lib/python3.5/site-packages/ipykernel_launcher.py:1: UserWarning: Boolean Series key will be reindexed to match DataFrame index.
  """Entry point for launching an IPython kernel.






    Out[22]:





Timestamp('1988-07-16 13:34:52.393093888')

Quel est le nombre moyen de paris par client ?



In [18]:

    
df_bc.groupby('UserId')['BetId'].count().mean()









    Out[18]:





104.50810985721797

Combien un client pari en moyenne ?



In [121]:

    
df_bc.groupby(['UserId', 'BetId'])['BetAmount'].sum().mean()









    Out[121]:





8.323373382295683

Combien un client gagne en moyenne ?



In [122]:

    
df_bc.groupby(['UserId', 'BetId'])['AmountWon'].sum().mean()









    Out[122]:





6.850836186417133

Montant des paris entré par le client ?



In [126]:

    
df_bc.drop_duplicates(subset='BetId')['BetAmount'].sum()









    Out[126]:





21505549.27

Montant gagné par le client ?



In [127]:

    
df_bc.drop_duplicates(subset='BetId')['AmountWon'].sum()









    Out[127]:





17700875.400000013

Montant gagné par betclic ?



In [125]:

    
df_bc.drop_duplicates(subset='BetId')['BetAmount'].sum() - df_bc.drop_duplicates(subset='BetId')['AmountWon'].sum()









    Out[125]:





3804673.869999986

Combien de paris par jours en moyenne ?



In [129]:

    
df_bc.groupby(['BetDate'])['BetId'].count().mean()









    Out[129]:





7059.437158469946

Les femmes gagnent-t-elles plus que les hommes ?



In [143]:

    
df_bc.groupby(['Gender', 'BetId'])['AmountWon'].sum().reset_index().groupby('Gender')['AmountWon'].mean()









    Out[143]:





Gender
F    6.513041
M    6.880524
Name: AmountWon, dtype: float64

Quel pays gagne le plus en proportion ?



In [152]:

    
df_bc.groupby(['Country', 'BetId'])['AmountWon'].sum().reset_index().groupby('Country')['AmountWon'].mean().sort_values(ascending=True).plot(kind='barh', figsize=(20, 5))









    Out[152]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8fa13af98>

Quel pays joue le plus ?



In [153]:

    
df_bc.groupby(['Country'])['BetId'].sum().sort_values(ascending=True).plot(kind='barh', figsize=(20, 5))









    Out[153]:





<matplotlib.axes._subplots.AxesSubplot at 0x7fe8fa201748>