NestedLogit.sample_posterior_predictive

NestedLogit.sample_posterior_predictive(choice_df=None, extend_idata=True, **kwargs)

Sample from posterior predictive distribution.

Parameters:

choice_dfpd.DataFrame, optional

New choice data for prediction. If None, uses training data.

extend_idatabool, optional

Whether to add to self.idata

**kwargs

Additional arguments for pm.sample_posterior_predictive

Returns:

az.InferenceData

Posterior predictive samples