Bivariate Poisson

[ckrapu]

Here’s an example notebook with a few ways to implement the correlated Poisson model. There’s more discussion on this thread which is essentially the same topic:

Dependent histogram count data

Hey! This is perhaps a broad question not limited specifically to PyMC, but I figured this is a good place to ask My input data is standard (as in, some vector of of covariates for each sample), but my observations are less so. In fact, my observations are a matrix (a vector for each sample) with histogram count data. Specifically, for each sample, the target is a vector of counts for each bin, which are of course somewhat correlated: index | X | bin_0 | bin_1 | bin_2 | ... | …

There’s a lot to unpack in that example notebook - feel free to call out specific parts that you’d like explained.