I have several set of probability distributions. I want to reliably estimate consistency across distributions inside each set.

Literature contains methods to compare two distributions: Kullback-Leibler divergence, earth mover's distance, ... But I couldn't find anything for more than two distributions.

I can imagine an indice built on top of one of these methods. For example, I could first merge all probability distributions into a single one. And then use the mean (or median) of the symmetric KL (or any other distance/divergence/...) between the merged distribution and all others distributions in the considered set. But before starting something out of the blue, I would rather be sure that nothing already exists.


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.