Is adstock transformation applied in budget scenarios?

[JJohnson-DA]

I've been looking through the code and I'm having a hard time tracking if adstock transformations are applied to the time periods when running sample_response_distribution.

I traced back through _create_synth_dataset, sample_posterior_predictive, and _data_setter and I only see the channel scaling transformation being applied.

Is this true? Is the scenario spend not adstocked?

[PabloRoque]

Hi @JJohnson-DA,

when calling sample_posterior_predictive we are generating forward samples of var_names=["y", "channel_contribution"].

In both the multidimensional and unidimensional MMMs channel_contribution is defined as the forward_pass (Adstock+Saturation, with order depending on your set-up) of the input variable.

So the answer is yes, the adstock is being applied.

[JJohnson-DA]

Perfect. Thank you!