predict_parameters

SuperLSS.predict_parameters(

X: object | EagerFrame,

*,

offsets: Mapping[str, NDArray] | None = None,

) → DataFrame

Return every fitted distribution parameter as a DataFrame.

Columns follow parameter_names_. Values are on each parameter’s natural scale after applying its inverse link. Gaussian columns are "location" and "scale"; Tweedie columns are "mean", "dispersion" and "power". Frame indices are preserved.

Supply prediction offsets by family parameter name when the model uses them. Their values are additions on the link scale.