Benjamin T. Vincent

At the moment `plot_posterior_predictive` has hard-wired HDI's. Plots look like this (taken from MMM docs page):  This is the relevant line: https://github.com/pymc-labs/pymc-marketing/blob/aa5fa1d280eae270a7a992dbc82f77586a02f237/pymc_marketing/mmm/base.py#L414 Can we...

The Bayesian approach to MMM is appealing because the posteriors over parameters encodes not only our beliefs about those parameters, but also how certain/uncertain we are about their values. If...

The `pymc-experimental` repo contains a [`ModelBuilder` class](https://www.pymc.io/projects/experimental/en/latest/generated/pymc_experimental.model_builder.ModelBuilder.html#pymc_experimental.model_builder.ModelBuilder) which can help in productionized / deployment situations because of the save/load functionality. It would be nice to bring this functionality in to...

We had a report of a spelling mistake "allot" (should be "a lot") in this notebook https://causalpy.readthedocs.io/en/latest/notebooks/iv_pymc.html Let's add a spell check as part of the pre-commit checks. There might...