Hi there,
I’m using the pymc mmm package. I understand control variables are not scaled by default, so I am manually scaling these. My input data table is therefore scaled control variables but unscaled channel variables (I believe pymc scales channel variables using maxabs scaling by default). I’ve been using functions such as plot_waterfall_components_decomposition in original scale. However pymc will assume the scaled control variables are the “original scale”. Does anyone know if there is a way round this? Is there a way to scale the control variables within the mmm model function?
Many thanks