What sort of analysis should be done in order to define our target outcome for modelling customer lapse?

I am trying to build a model to predict customer lapse and am required to define the target lapse definition myself.

What sort of customer behavioural analysis should I do in order to define my modelling target? In other words, what sort of analysis could help me decide between a lapse definition of 90 days without a purchase compared to 180 days.

Topic churn predictive-modeling

Category Data Science

1
answered at

If you choose to do the analysis in a scripting language, you can define days_since_last_purchase as a variable. Then you can systematically change the variable to see which value makes the most sense.

There are a couple of options:

  • Look for an inflection point in the performance.
  • Show the results to a domain expert to choose a target value.
  • Use a distribution of possible target values instead of a single point estimate, aka a Bayesian perspective.

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.