Will Dean

Results 623 comments of Will Dean

Yeah, thats why I have the second item to add documentation for each transformation class

What is the status here?

What is the action here? Add this to the documentation? Or just have this as an issue to reference?

Hi @kb-open, this is an open source project. Would you like to make this contribution?

not sure how eval would work in the case where shapes are not provided. ```python a = pt.vector("a", dims="channel") b = pt.vector("b", dims="geo") result = a + b # raise...

> There's also the question of what should be the output, xarrays? Because if dims order is arbitrary users don't know what they're getting, but building xarray for the output...

I am against adding the `hdi_list` kwargs to `sample_prior_predictive` method `sample_prior_predictive` returns the InferenceData object. The user would be able to call az.hdi on this object. More control is given...

You can initialize as a constant value, say 1.0. Maybe the deserialization doesn't work though

It and sweep_values seem to be positional only: https://github.com/pymc-labs/pymc-marketing/blob/eb2f846d42139c55ec3e2ae3689536839f7781ee/pymc_marketing/mmm/sensitivity_analysis.py?plain=1#L289-L293 Would you like to make the fix in the documentation, @nipnipj?