I am currently looking into the application of mutual information analysis for data produced in our laboratory. This stated, I was curious if there are any quantitative methods for determining the 'correct' number of bins for the histograms used to compute the mutual information between two data sets. As I understand it, the computation of mutual information can be heavily dependent on the number of bins chosen for the computation, and in my own preliminary investigations, this has been confirmed. Currently, using the CRP toolbox by Marwan, I was wondering if anyone had any suggestions for determining the optimal binning.
Thank you in advance.
Regards,
Kris