Fixing the shape argument (and associated comments) for the observed multinomial draws of votes (correct shape is num_precincts x num_columns) in the r by c models. Note: I'm not totally sure why this argument is necessary, since this is the observed variable (so has access to the shape of the observations), and it seemed to get ignored in previous runs. However, omitting the shape argument entirely seems to leads to errors when using the latest version of pymc3 (but not in version 3.9.3). This PR also updates the r by c tests to include an example with r not equal to c.
Fixes #91
Fixing the shape argument (and associated comments) for the observed multinomial draws of votes (correct shape is num_precincts x num_columns) in the r by c models. Note: I'm not totally sure why this argument is necessary, since this is the observed variable (so has access to the shape of the observations), and it seemed to get ignored in previous runs. However, omitting the shape argument entirely seems to leads to errors when using the latest version of pymc3 (but not in version 3.9.3). This PR also updates the r by c tests to include an example with r not equal to c.