Geo vs Hierarchical level Model insights

[Gangesh-C]

Hi Team,

I am trying to built an Hierarchical MMM model for different geo using PyMC. I am able to build an hierarchical structure and generate posterior prediction. However, I am stuck on how to showcase different insight for each geo and overall level:

Insights like:

1. Channel contribution to revenue (waterfall kind of chart)
2. Saturation curve
3. How to fit budget allocator for this?

Can you guys share some direction or references, which we can refer here ?

Data: 5 spend channels with Revenue as KPI, at different geo level (4 currently)

Basic model structure is:
with pm.Model() as bhmmm:
--- parameters and other functionalities
trace = pm.sample()

Note: New to this kind of modelling , let me know if I should share any other info.

[williambdean]

Hi, @Gangesh-C

You might be interested in this PR #1521 which builds upon our new pymc_marketing.mmm.multidimensional.MMM class. This class allows for specifying extra dimensions in the model. The model will be hierarchical depending upon the priors that are specified. For example

from pymc_marketing.prior import Prior

independent_prior = Prior("Normal", mu=Prior("Normal"), sigma=Prior("HalfNormal"), dims=("channel", "geo"))
hierarchical_prior = Prior("Normal", mu=Prior("Normal", dims="channel"), sigma=Prior("HalfNormal", dims="channel"), dims=("channel", "geo"))

model_config = {
    "saturation_...": independent_prior,
}

In short, the configuration you use defines the structure of the model.

Important

This is a new experimental class so proceed with caution. But we are open to feedback and can help you with any issues you might face.