DistRegBuilder.add_predictor()

DistRegBuilder.add_predictor(name, inverse_link)

Adds a predictor to the model builder.

Parameters:

• name (str) – The name of the parameter of the response distribution. Must match the name of the parameter of the TFP distribution.

• inverse_link (type[Union[Bijector, Bijector]]) – The inverse link mapping the regression predictor to the parameter of the response distribution. Either a string identifying a TFP bijector, or alternatively, a TFP-compatible bijector class. If a class is provided instead of a string, the user needs to make sure it uses the right NumPy implementation.

Return type:

DistRegBuilder