I'm working on a classical churn prediction problem using the number of visits of a given user to a site and I thought that Poisson Regression was the right tool for modelling the future engagement of that user. When then I came across a book about survival analysis and Hazard Modelling and I don't know which technique is best.

I don't want to be researching both topics at the same time, so what is best for modelling user engagement using past data and demographics?


Brief and general answer:

  • With Poisson regression, the response variable of interest is a count (or possibly a rate).
  • With Cox regression (or alternative modelling strategies from survival analysis), the response variable is the time that has elapsed between some origin and an event of interest. In particular, survival analysis techniques are designed to handle censoring.
  • Note that, under some assumptions, there is a link between the two.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.