I’m a PyMC3 newbie. I’m trying to create a Bayesian Rasch model, but I don’t have all the observations (i.e. an observation for every user, for every question). I understand that ordinarily, for missing observations PyMC3 automatically performs imputation by sampling from the prior. In my case howev…

Dealing with missing data and custom distribution

ricardoV94 March 12, 2021, 5:44pm 10

The logp shouldn’t see the missing values, it should see whatever values are proposed by the NUTS inplace of the missing values.

Topic		Replies	Views
Missing values in a model? Questions	12	4791	November 7, 2018
What does the hierarchical model look like when having missing in observed? Questions	12	1173	October 31, 2018
Rewriting Likelihood with DensityDist() outputs MissingInputError Questions	9	848	April 3, 2021
Marginalizing over missing categories Questions	1	726	June 17, 2020
Calculating the likelihood based on not missing observed values Questions	5	1995	June 12, 2018