PyMC+ArviZ: how to make the most of labeled coords and dims in PyMC 4.0

I had a quick go of trying to directly pass a multi-index as a dim in a use-case where a >2 level model might be justified, in a geographic hierarchy. It seems like there is a hiccup in the dimension checking, though. Here is the example notebook with the dimension error.

I’ll try to muddle though, but hoping you can quickly see something I won’t. It seems to be a conflict between the fact that the array representation is only 2d, but I want to label 5 dimensions.

1 Like