Will Dean

Brainstorming how this might look ``` y ~ 1 + menten(geometric(Google + Facebook, max_lag=8)) ``` Guess each channel would be specified inside parans? Would some channels get some transformations and...

I have a [PR](https://github.com/pymc-labs/pymc-marketing/pull/344) to add ax argument to the current CLV plotting functions. Might be a good standard here in order to give additional flexibility to the user

Love that idea. Separate the data transformation from plotting. Total sense to me

> I tried to find a `sample_posterior_predictive` function for the `BetaGeoModel` for comparison, but I'm having trouble getting the one from pymc to work. Is there are standard way of...

Can you share a bit more about what you were running and how you were running it? It looks like a jupyter notebook error message. Are you able to run...

Hi @rym903 My first thought at the moment would be to: - fit model to get posterior - build model again with these new spends - sample for "channel_contributions" with...

Did you set the extend = True when using the 'sample_posterior_predictive' method?

Awesome. Yeah, It doesn't seem too hard of an adaption. Guessing without a guard, the `orelse` attribute of the `ast.If` will be the rest of the `body` in the case...