Facebook-C1.py

# coding: utf-8

# In[1]:

import pandas as pd
import numpy as np
from sklearn import ensemble, feature_extraction, preprocessing
from IPython.core.display import HTML

bids = pd.read_csv('/home/alfard/Facebook-Robot/bids.csv')


# In[2]:

#bids.to_pickle(('/home/alfard/Facebook-Robot/test.pk'))

#bids=pd.read_pickle(('/home/alfard/Facebook-Robot/test.pk'))


# In[3]:

#jour 1
#Bids dans 1
bids2=bids
bids2['time2'] = bids2['time']/100000000000

bids_D1=bids2[(bids2['time2'] > 96318) & (bids2['time2'] < 96456)]

bids_D2=bids2[(bids2['time2'] > 96954) & (bids2['time2'] < 97093)]

bids_D3=bids2[(bids2['time2'] > 97591) & (bids2['time2'] < 97729)]

del bids


# In[4]:

#Nombre de bids dans la journée 1

D1_bid=bids_D1[['bidder_id', 'bid_id']].groupby(['bidder_id']).agg(['count'])


# In[5]:

#D1_bid nombre de bid dans une journée
#Index en valeur
D1_bid['bidder_id'] = D1_bid.index

D1_bid.columns=['D1_bid','bidder_id']

D1_bid = D1_bid[['bidder_id','D1_bid']]

#D1_bid ok
#display(HTML(D1_bid.to_html()))


# In[6]:

#D1_action Nombre d'action at the same time
nbact=bids_D1[['bidder_id', 'auction']]
nbact.shape

D1_auction=nbact.groupby(nbact['bidder_id'])['auction'].apply(lambda x: len(x.unique()))

D1_auction=pd.DataFrame(D1_auction)

D1_auction['bidder_id'] = D1_auction.index

D1_auction.columns=['D1_auction','bidder_id']

D1_auction=D1_auction[['bidder_id','D1_auction']]


# In[7]:

#D1 standart deviation
D1_std=bids_D1[['bidder_id','time2']].groupby(['bidder_id']).agg(['std'])

D1_std['bidder_id'] = D1_std.index

D1_std.columns=['D1_std','bidder_id']

D1_std = D1_std[['bidder_id','D1_std']]


# In[8]:

print D1_bid.shape
D1_bid.to_pickle(('/home/alfard/Facebook-Robot/D1_bid.pk'))

print D1_auction.shape
D1_auction.to_pickle(('/home/alfard/Facebook-Robot/D1_auction.pk'))

print D1_std.shape
D1_std.to_pickle(('/home/alfard/Facebook-Robot/D1_std.pk'))


# In[9]:

#DAY 2

D2_bid=bids_D2[['bidder_id', 'bid_id']].groupby(['bidder_id']).agg(['count'])

D2_bid['bidder_id'] = D2_bid.index

D2_bid.columns=['D2_bid','bidder_id']

D2_bid = D2_bid[['bidder_id','D2_bid']]

#D2_action Nombre d'action at the same time
nbact=bids_D2[['bidder_id', 'auction']]
nbact.shape

D2_auction=nbact.groupby(nbact['bidder_id'])['auction'].apply(lambda x: len(x.unique()))

D2_auction=pd.DataFrame(D2_auction)

D2_auction['bidder_id'] = D2_auction.index

D2_auction.columns=['D2_auction','bidder_id']

D2_auction=D2_auction[['bidder_id','D2_auction']]


#D2 standart deviation
D2_std=bids_D2[['bidder_id','time2']].groupby(['bidder_id']).agg(['std'])

D2_std['bidder_id'] = D2_std.index

D2_std.columns=['D2_std','bidder_id']

D2_std = D2_std[['bidder_id','D2_std']]


# In[10]:

print D2_bid.shape
D2_bid.to_pickle(('/home/alfard/Facebook-Robot/D2_bid.pk'))

print D2_auction.shape
D2_auction.to_pickle(('/home/alfard/Facebook-Robot/D2_auction.pk'))

print D2_std.shape
D2_std.to_pickle(('/home/alfard/Facebook-Robot/D2_std.pk'))


# In[11]:

#DAY 3

D3_bid=bids_D3[['bidder_id', 'bid_id']].groupby(['bidder_id']).agg(['count'])

D3_bid['bidder_id'] = D3_bid.index

D3_bid.columns=['D3_bid','bidder_id']

D3_bid = D3_bid[['bidder_id','D3_bid']]

#D3_action Nombre d'action at the same time
nbact=bids_D3[['bidder_id', 'auction']]
nbact.shape

D3_auction=nbact.groupby(nbact['bidder_id'])['auction'].apply(lambda x: len(x.unique()))

D3_auction=pd.DataFrame(D3_auction)

D3_auction['bidder_id'] = D3_auction.index

D3_auction.columns=['D3_auction','bidder_id']

D3_auction=D3_auction[['bidder_id','D3_auction']]


#D3 standart deviation
D3_std=bids_D3[['bidder_id','time2']].groupby(['bidder_id']).agg(['std'])

D3_std['bidder_id'] = D3_std.index

D3_std.columns=['D3_std','bidder_id']

D3_std = D3_std[['bidder_id','D3_std']]


# In[12]:

print D3_bid.shape
D3_bid.to_pickle(('/home/alfard/Facebook-Robot/D3_bid.pk'))

print D3_auction.shape
D3_auction.to_pickle(('/home/alfard/Facebook-Robot/D3_auction.pk'))

print D3_std.shape
D3_std.to_pickle(('/home/alfard/Facebook-Robot/D3_std.pk'))


# In[14]:

#CLEAN
del bids_D1,bids_D2,bids_D3,nbact


# In[2]:

#COUNTRY DUMMIES

country=bids[['bidder_id','country']]

country_1 = pd.get_dummies(country['country'])

toto=bids[['bidder_id']]

#country_2=pd.concat([toto,country_1],axis=1)


# In[3]:

del bids


# In[4]:

country_3=pd.concat([toto,country_1],axis=1)


# #display(HTML(T1.head(5).to_html()))
# http://stackoverflow.com/questions/25145317/pandas-merge-two-dataframes-with-identical-column-names

# In[5]:

del country_1,toto


# In[6]:

country_4=country_3.groupby(['bidder_id']).agg(['sum'])


# In[7]:

del country_3

country_4.to_pickle(('/home/alfard/Facebook-Robot/country_4.pk'))

#country_4['bidder_id'] = country_4.index
#country_B=country_4.rename(columns = {('bidder_id', ''):'bidder_id'})

#del country_4

#country_B.shape


# ***********************************************************************************

# In[10]:

country_4['bidder_id'] = country_4.index
list(country_4)


# In[13]:

country_B=country_4.rename(columns = {('bidder_id', ''):'bidder_id'})
country_B=country_B.rename(columns = {('bidder_id', ''):'bidder_id'})
list(country_B)


# In[16]:

country_B['bidder_id'].head()


# In[2]:

#MERCHANDISES DUMMIES

merch=bids[['bidder_id','merchandise']]

merch_1 = pd.get_dummies(merch['merchandise'])

toto=bids[['bidder_id']]

merch_2=pd.concat([toto,merch_1],axis=1)


# In[3]:

merch_3=merch_2.groupby(['bidder_id']).agg(['sum'])


# In[4]:

merch_3['bidder_id'] = merch_3.index
merch_B=merch_3.rename(columns = {('bidder_id', ''):'bidder_id'})

del merch,merch_1,toto,merch_2,merch_3 

merch_B.to_pickle(('/home/alfard/Facebook-Robot/merch_B.pk'))

merch_B.shape


# *******************************************************************************************

# In[6]:

#DEVICE DISTINCT
nbdev=bids[['bidder_id', 'device']]
nbdev.shape

device_NB=nbdev.groupby(nbdev['bidder_id'])['device'].apply(lambda x: len(x.unique()))

device_NB=pd.DataFrame(device_NB)

device_NB['bidder_id'] = device_NB.index

device_NB.columns=['device_NB','bidder_id']

device_NB=device_NB[['bidder_id','device_NB']]

device_NB.to_pickle(('/home/alfard/Facebook-Robot/device_NB.pk'))

device_NB.shape


# **************************************************************************

# In[8]:

#IP DISTINCT
nbip=bids[['bidder_id', 'ip']]
nbip.shape

ip_NB=nbip.groupby(nbip['bidder_id'])['ip'].apply(lambda x: len(x.unique()))

ip_NB=pd.DataFrame(ip_NB)

ip_NB['bidder_id'] = ip_NB.index

ip_NB.columns=['ip_NB','bidder_id']

ip_NB=ip_NB[['bidder_id','ip_NB']]

ip_NB.to_pickle(('/home/alfard/Facebook-Robot/ip_NB.pk'))

ip_NB.shape


# **********************************************************************

# In[10]:

#URL DISTINCT
nburl=bids[['bidder_id', 'url']]
nburl.shape

url_NB=nburl.groupby(nburl['bidder_id'])['url'].apply(lambda x: len(x.unique()))

url_NB=pd.DataFrame(url_NB)

url_NB['bidder_id'] = url_NB.index

url_NB.columns=['url_NB','bidder_id']

url_NB=url_NB[['bidder_id','url_NB']]

url_NB.to_pickle(('/home/alfard/Facebook-Robot/url_NB.pk'))

url_NB.shape


# **********************************************************

# In[ ]:

#Merge all table

#D1_bid
#D1_auction
#D1_std

#D2_bid
#D2_auction
#D2_std

#D3_bid
#D3_auction
#D3_std

#country_B
#merch_B

#device_NB
#ip_NB
#url_NB