I have two KDE distributions, each with a different number of bins. I'd like to compare them effectively, and I'm wondering if there's a recommended technique for this. Should I unify the number of bins, and if so, what's the best approach?

  • 1
    $\begingroup$ How do you like to compare the distributions? Like compute some sort of distance? Which metric do you use and why can it not be applied to your estimated distributions? $\endgroup$ Jan 16 at 6:36
  • $\begingroup$ @SextusEmpiricus yes, I mean to calculate the distance between the KDEs $\endgroup$ Jan 16 at 9:10
  • 1
    $\begingroup$ Which distance? $\endgroup$ Jan 16 at 9:46
  • $\begingroup$ How are estimates defined? With bins creating a stepwise function looking like a histogram or as a continuous function? $\endgroup$ Jan 16 at 9:47
  • $\begingroup$ @SextusEmpiricus discontinue function. $\endgroup$ Jan 16 at 9:48


Your Answer

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