Churn prediction on BPI data example code request

[burakbayramli]

I'd like to share with you the BPI customer data I received from the author WTTI-RNN

Data

This could make an interesting use case of your code. The way to connect the data is:

import pandas as pd, zipfile, os
with zipfile.ZipFile('bpi_visit_churn.zip', 'r') as z:
    df =  pd.read_csv(z.open('bpi_visit.csv'),sep='\s')
df2 = df[df['cid'] == 1076745]
print len(df2)
print df2.tail(10)

It shows

           cid        date
11528  1076745  2015-11-25
11529  1076745  2015-12-02
11530  1076745  2015-12-03
11531  1076745  2015-12-08
11532  1076745  2015-12-09
11533  1076745  2015-12-31

So per customer we get a sorted list of dates, which can be seen as visits. Churn could be used to predict a customer's loyalty.