pymc-marketing

pymc-marketing copied to clipboard

Discuss `fit()` Conventions

1

What should be returned when `fit` is called? `fitted_model = Model(data).fit()` is a common convention amongst those familiar with the `sklearn` API, but won't work unless `self` is returned. Currently...

ColtAllen

MMM mixins as single sklearn transformer

2

The sklearn transformers that require state (MaxAbsScaler and StandardScaler) are currently part of the mixins which make it a bit difficult to support new data. So the mixins will have...

wd60622

Allow specifying prior distribution in MMM models via config

4

https://github.com/pymc-labs/pymc-marketing/blob/d4feeeafad6f0299b115dac3582ad6d98bd8241f/pymc_marketing/mmm/delayed_saturated_mmm.py#L123-L126 Right now it is mixed logic of creating RV and registering it in the model. In principle that should be the only one method that returns already registered RV....

ferrine

Use bambi for MMMs

5

ricardoV94

Avoid PyTensor evals whenever possible

3

In `BaseDelayedSaturatedMMM`, the `channel_contributions_forward_pass()` method calls `eval()` on the pytensor value to be returned ([here](https://github.com/pymc-labs/pymc-marketing/blob/3a9db517d1d3c5a930eeb80ad24c540382d1bc33/pymc_marketing/mmm/delayed_saturated_mmm.py#L386)). This causes problems if you expect the tensor returned by the various transformation methods to...

cluhmann

Add optimization notebook

8

Initial version of the optimization notebook. I have not yet added all the "bells and whistles" that are possible, because I want to get feedback on the core first. Then...

cluhmann

Add `kwargs` to `fit` method in `ModelBuilder` for prior and posterior predictive sampling steps.

2

At the moment the `.fit` methods does 3 samplings: prior predictive, inference and posterior predictive. At the moment, we can just pass `kwargs` to the sampler but not to the...

juanitorduz

List of Continuous, Non-Contractual `clv` Models to Consider Adding

10

https://github.com/mplatzer/BTYDplus

ricardoV94

Please can you add Hierarchical model for MMM

2

kurvaraviteja355

Implement validation PPC plots for CLV models

1

https://lifetimes.readthedocs.io/en/latest/Quickstart.html TODO: - Find out what metrics we want to do PPC wrt, implement plots - Check if there are model-specific ones as an extra (look at lifetimes for examples)

ricardoV94