ExperimentDesigner.from_idata

classmethod ExperimentDesigner.from_idata(idata, saturation='logistic', adstock='geometric')

Create an ExperimentDesigner from a saved InferenceData fixture.

This constructor builds a lightweight PyMC model using the specified transformation classes, then compiles a graph-based evaluation function — the same approach as the main constructor.

Parameters:

idataInferenceData

An ArviZ InferenceData containing posterior samples with variables saturation_lam, saturation_beta, and adstock_alpha (with a channel dimension), plus constant_data with current_weekly_spend, residual_std, l_max, and normalize.

saturationstr

Saturation function type. Currently only "logistic" is supported.

adstockstr

Adstock function type. Currently only "geometric" is supported.

Returns:

ExperimentDesigner

A configured designer ready for recommend().