MaxDiffMixedLogit.make_model

MaxDiffMixedLogit.make_model(arrays, observed=True)

Build the MaxDiff PyMC model.

Parameters:

arraysMaxDiffArrays

Preprocessed padded-plus-mask representation of the tasks.

observedbool, default True

Whether to attach observed data to the Categorical likelihoods. When False, the model can be used for forward simulation.

Returns:

pm.Model

Model with beta_item/beta_item_r (intercept mode) or beta_feat/U_item_r (part-worths mode) plus two observed Categorical likelihoods best_pick / worst_pick.