Will Dean

I think that is a fine location. I'd say, throw it directly in a new submodule of `pymc_marketing.mmm` as well. If you want to make it harder to reach then...

Awesome talk, @drbenvincent ! Would this just require changing the likelihood for the model? How would the saturation parameters be affected by this?

Might be related to this new doc showing linear effects can be better estimated via use of lift tests: https://www.pymc-marketing.io/en/latest/notebooks/mmm/mmm_roas.html What are you assumptions of the model? CC: @juanitorduz

Ref to what fails when no pymc cap @juanitorduz @ColtAllen `make_node` method seems to be main issue. The tests with oldest supported pymc werent run due to auto stopping but...

> Thanks @wd60622 ! I think we could tackle this issue once 5.17 is released and support from 5.16 onwards (so that we support two versions) WDYT? Probably the easiest....

> > > Thanks @wd60622 ! I think we could tackle this issue once 5.17 is released and support from 5.16 onwards (so that we support two versions) WDYT? >...

The `combined_dims` property would likely be affected with a change

A downside of this is shows up with https://github.com/pymc-labs/pymc-marketing/issues/1693 https://github.com/pymc-labs/pymc-marketing/issues/1703 Since the deterministics that are created are just broadcastable instead of having the dims. The behavior of that workflow would...

Maybe time since first purchase? Would you know @ColtAllen?

The most similar logic is [`EventAddivitiveEffect`](https://www.pymc-marketing.io/en/latest/api/generated/pymc_marketing.mmm.additive_effect.EventAdditiveEffect.html) while using the multidimensional MMM class. Users can define custom basis function to their liking. Reference: https://www.pymc-marketing.io/en/latest/api/generated/pymc_marketing.mmm.events.html Different priors will likely require adding separately...