Bayesian newbie here. I have a 4x4 data matrix consisting of proportions of success (probability of observed successes) and would like to obtain parameters alpha (4-vector for rows), beta (4-vector for columns) and c using MCMC such that the posterior distribution in data[i,j] is given by 1/(1 + exp…

Model not reproducing observed data?

junpenglao July 24, 2018, 6:42pm 2

You cannot feed number other than [0, 1] to the Bernoulli distribution, it probably cast data to the closest integer internally.

Topic		Replies	Views
Samples from prior appear to have wrong distribution Questions	2	606	October 8, 2018
Likelihood given by product of two distributions? Questions	6	2215	August 4, 2018
Hierarchical bernoulli model with bernoulli priors Questions	1	403	May 1, 2020
Bernoulli samples all the same? Questions	1	2449	May 29, 2018
Problem model definition PyMC3 Questions	12	4749	September 21, 2017