1
$\begingroup$

I have a dataset of inactive users for the last 365 days, with columns:

  1. When subscribed (e.g. 10.10.2022)
  2. When unsubscribed (e.g. 12.10.2022)

and client info

I also have a set of data about active users:

  1. When subscribed

and client info.

Question: How can I get a prediction of whether a certain user will unsubscribe within a month from now? What should be the logic of my actions in this case?

I tried to calculate an unsubscribe date using linear regression and got a result, for example, 03.15.2023. But in that case, I can't understand with what probability the person will unsubscribe in a month, namely on 01.12.2023

I will be grateful for any hints. Thanks

$\endgroup$
3
  • $\begingroup$ You will first need to preprocess your data by converting the "when subscribed" and "when unsubscribed" columns to a numerical format that the model can understand. $\endgroup$
    – Vic
    Dec 12, 2022 at 13:40
  • $\begingroup$ @Vic Yes, I converted them to the Unix timestamp format. There is no problem with that. But thanks for the advice. $\endgroup$
    – Quaw
    Dec 12, 2022 at 13:47
  • $\begingroup$ You may be interested in survival analysis. $\endgroup$
    – Andy W
    Dec 12, 2022 at 14:13

1 Answer 1

0
$\begingroup$

This problem is better framed as survival analysis, the expected time duration until an event occurs. The goal would be to fit a model that predicts the probability of churn as a function of the number of days thus far and other features.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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