density_fan

SuperLSS.density_fan(

reference: Mapping[str, Any] | Series,

covariate: str,

*,

X_train: object | EagerFrame | None = None,

**kwargs: Any,

) → DensityFan

Return the conditional response density along one covariate.

The same sweep as risk_curves(), but the whole density at each point rather than a few quantiles of it: it is the picture that shows a shape change – a mass moving into the tail, a mode splitting – which no set of quantile curves states outright. This payload supports continuous families only; families with atoms refuse. weights gives positive prior-law weights per swept point, with the unit law as the default, as in risk_curves().