How to avoid collapsing to one view?

probcomp / crosscat

A domain-general, Bayesian method for analyzing high-dimensional data tables

http://probcomp.csail.mit.edu/crosscat/

Apache License 2.0

322 stars 42 forks source link

How to avoid collapsing to one view? #115

Open kayhan-batmanghelich opened 7 years ago

kayhan-batmanghelich commented 7 years ago

Hi,

I was wondering where you set the hyper-parameters of the CRP ? I ran crosscat on my data and it always collapses to one view. The logscore also converges too quickly (about 10 transitions) which is not reasonable for a sampling approach. What parameters do you suggest to tune?

Thanks,