Dato Churn Prediction Toolkit

User 5331 | 6/26/2016, 5:13:29 PM

Hello , I have a customers transactions dataset in which I have WEEK_NO variable with values {01,02.....102} with 102 weeks. I want to find out customer churn per week.Problem I am facing is I dont have a 'datetime' variable as mentioned in graphlab churn prediction blog.How can I proceed?


User 16 | 6/27/2016, 8:46:17 PM

I suggest creating a time column, using the WEEKNO variable. Set rows with WEEKNO value of 01 to what ever the starting time was. Then set values of 02 to one week later. This should be easily to do using the timedelta function in the datetime module, that comes standard with Python.