I understand the the cumulative gains chart shows the percentage of found targets (y-axis) (so the sensitivity) against the number of tested observations (x-axis).

For the lift chart it seems that the y-axis shows the improvement for the percentage of found targets compared to the hypothetical random model which is a line from 0-100% in the cumulative gains chart, this is called lift. However in some explanations, the lift is still plotted against the number of tested observations,

enter image description here

in other explanations the lift is plotted against a "cut-off" (whatever that means),

enter image description here

(obviously there is something wrong with the curves itself, but this is a programming issue that I produced myself - I think the curves need to be flipped horizontally). In others the lift is plotted against the "rate of positive predictions".

enter image description here

I guess the "rate of positive predictions" means the same as true positive rate a.k.a. sensitivity a.k.a. recall a.k.a. hit rate. What is correct for a lift chart?

  • $\begingroup$ Could you display the charts you're asking about? Not everyone uses the same names to refer to the same types of charts. Or graphs. Or plots. $\endgroup$ – rolando2 Jan 11 at 18:56
  • $\begingroup$ @rolando2: Not sure how this helps, but here you go. $\endgroup$ – Make42 Jan 11 at 22:09
  • $\begingroup$ Have you read this paper? $\endgroup$ – Lerner Zhang May 22 at 9:25

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.