Does it? I'm not exactly certain. I fit two KDE PDFs on two datasets. I want to measure the disimilarity between them.
Asked
Active
Viewed 1,914 times
3
-
5Kl divergence (and other information theoretic quantities) can be estimated directly from the data rather than through KDEs. See e.g. http://www.tsc.uc3m.es/~fernando/bare_conf3.pdf – Georg M. Goerg Jul 26 '16 at 17:10
-
What if the sets differ in size? – Jul 26 '16 at 17:28
-
In the paper the number of samples is different, n and m (see first sentence of section 2). – Georg M. Goerg Jul 26 '16 at 17:50
-
cool; do you know if this is available in Python? – Jul 26 '16 at 18:14
-
1something like http://www.isi.edu/~gregv/npeet.html – Georg M. Goerg Jul 26 '16 at 22:38