MNLogit.apply_intervention

MNLogit.apply_intervention(new_choice_df, new_utility_equations=None)

Apply one of two types of intervention.

The first type of intervention assumes we have a fitted model and just aims to sample from the posterior predictive distribution after adjusting one of more of the models observable attributes and passing in the new_choice_df. The second type of intervention allows that we remove a product entirely from the market place and model the market share which accrues to each product in the adjusted market.