GPBoost
GPBoost copied to clipboard
Feature Request: Multivariate modeling via coregionalization
I am interested in an extension to joint modeling of multiple correlated outcomes. For example, via coregionalization, described in this paper and implemented in pymc3. I see a separable spatiotemporal GP on your roadmap, and I think the implementation of coregionalization would be similar.
Nice work, BTW. I recently came across this software and the two supporting papers and am excited to apply these methods to my work.
Thank you for your interest in GPBoost.
I currently do not have time to work on this, but I will add it to the to-do list.