Help with Multivariate Normal with nested prior information

You are absolutely right! I must admit I thought I was increasing the iterations by increasing my samples on the fit method but yes fit was just defaulting out to 10k and not converging… Running perfectly now! pm.sample() seems to get to the right answer way faster (even with the default 1k). I had wrongly assumed fit method is always faster.

Learning so much, thanks to PyMC3.

Thanks again!

1 Like