Will Dean

Results 58 comments of Will Dean

Yes, I've noticed this too in a spots. Usually fixed by adding a space the docstring in order to separate the sections better. There might be a way to check...

Running list: ## [BaseMMM](https://www.pymc-marketing.io/en/stable/api/generated/pymc_marketing.mmm.base.BaseMMM.html#pymc_marketing.mmm.base.BaseMMM) - [ ] BaseMMM.compute_channel_contribution_original_scale - [ ] BaseMMM.get_target_transformer - [ ] BaseMMM.plot_channel_contribution_share_hdi - [ ] BaseMMM.plot_channel_parameter - [ ] BaseMMM.plot_component_contributions - [ ] BaseMMM.plot_prior_predictive ## [MMM](https://www.pymc-marketing.io/en/stable/api/generated/pymc_marketing.mmm.base.MMM.html#pymc_marketing.mmm.base.MMM)...

lambda is a keyword in python so it isnt typically used. https://www.programiz.com/python-programming/keyword-list I dont think think you can override it. Like with return or def In my mind, lam =...

Personally, I am for sklearn especially with transformation of dataframe X in order to support new data. The transformation that are currently used carry a state (max value or mean,...

> > Are there concerns of it as a dependency? > > The issue is that we can't integrate sklearn stuff with other PyMC/PyTensor methods, e.g., the optimizer stuff that...

Obviously that is the intended behavior of pytensor. However, I am talking within the context of a MMM. if `y` was `cost` then the max cost that is used to...

I am not attached to sklearn. However, I am for exchangeability of transformation methods. Having them all as mixins makes it almost very hard to change without having to rewrite...

I have [this PR](https://github.com/pymc-labs/pymc-marketing/pull/405/files) up which removes the mixins for preprocessing and uses a single transformer (ColumnTransformer) to handle the transformations of new data. It addresses a lot but the...

might also be a bad tests on the values. But definitely having a seed will have reproducible result Might be resolved with #515

Might be good to add the curve to the transformers.py file as an early stage and can incorporate it into the larger class later on