How to set bin size for computing mutual information from continuous variable?

Question

I want to compute the mutual information between features and my output variable. I was wondering, what is the best way to set the number of bins for each feature? How should be bins interval ? Does all of them should have same bin size? Here is how my data looks like:

 A         B            C          D       E
 0.30890524 0.54426331 0.100881953 1.7844281 1.9580541
 0.1037904 0.10647233 0.102095382 0.8488240 0.7763768
 0.10367904 0.147233 0.102095382 0.8488240 0.7763768
 0.331458 0.57973406 0.130334158 1.6350764 1.5585344
 0.15101780 0.2377797 0.150907454 0.8556408 1.0199345
 0.14075664 0.04942940 0.0103453 0.7010386 0.523710
6 0.02547862 0.01841224 0.04950307 0.1694650 0.1293436
 0.31318298 0.123281 0.387902840 1.013703 0.9320006
 0.545757 0.5898526 0.242701313 1.8950583 2.1294465
 0.332576 0.44881516 0.181627835 1.5116738 1.5444081

Please don't discretize continuous variables. (And many other threads on this site.) — Stephan Kolassa, Oct 01 '15 at 16:28
then how can I compute mutual information if I don't discritize it ? — user51661, Oct 01 '15 at 17:33
See https://stats.stackexchange.com/questions/391123/recommended-mutual-information-estimator-for-continuous-variable, https://stats.stackexchange.com/questions/147401/estimating-mutual-information-using-r, https://www.stat.berkeley.edu/~binyu/summer08/L2P2.pdf — kjetil b halvorsen, Aug 01 '20 at 19:01

How to set bin size for computing mutual information from continuous variable?

0 Answers0