I am trying to compute the uplift for some campaign. For the same I am building model/models. I need to know how much individual model performance should impact my uplift computation? Is there any relation between the two?

In simple words, if x percentage of error occurs in model predictions what percentage of error it will reflect in uplift computation?


This question has an open bounty worth +50 reputation from Artiga ending in 6 hours.

This question has not received enough attention.

  • $\begingroup$ How are you going to perform the uplift computations? Based on historic data? Where do the model/models come in action? I didnt understand what is it you are asking. $\endgroup$ – yoav_aaa Aug 15 at 9:57
  • $\begingroup$ What kind of model are you using -- in statistical terms? Maybe you're asking the relationship betw. coefficients and (rsq or standard errors of estimate or ...)? $\endgroup$ – rolando2 Aug 15 at 11:18
  • $\begingroup$ @rolando2 I'm using a simple Decision Tree model. Let me simplify again what I am asking. Let's say I have a model which has accuracy of 80% and hence 20% of error. So how much of this 20% of error will constitute to error in uplift computation? $\endgroup$ – Artiga Aug 16 at 4:41
  • $\begingroup$ @yoav_aaa, it is standard procedure of uplift modeling. First fit the model by modelling the persuadables and then compute the uplift. Just in case if you don't know about the procedure: medium.com/datadriveninvestor/… $\endgroup$ – Artiga Aug 16 at 4:46
  • $\begingroup$ @rolando2 No! I am not asking for the relationship between coefficients and R-squared. $\endgroup$ – Artiga Aug 16 at 4:56

Your Answer

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

Browse other questions tagged or ask your own question.