08-05-2014 10:43 PM
I have two questions about predictive model which need to seek for your advice!
We have a predictive model and pilot shows that the uplift is 45% which is statistically significant.
Now, we use it to generate targeting call center lead of top10% every month and those customers who are contacted cannot be contacted again in the next 5 months to avoid any complain.
Here is the arguement. Since those good customers are kept out of the population for the next 5 months, the uplift of top10% in the next few months will be depreciated assuming there is no new customers enter the population. Is it true? How to prove it?
Size of top 10% is 40,000 and monthly lead is 15,000.
2. How long can a robust predictive model sustain its uplift?
Any comment is highly appreciated!