plot_diagnostics

SuperLSS.plot_diagnostics(

X: object | EagerFrame,

y: NDArray,

*,

engine: str = 'matplotlib',

n_sim: int = 100,

max_points: int = 50000,

seed: int = 42,

sample_weight: NDArray | None = None,

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

) → Any

Draw the six-panel distributional diagnostic.

Q-Q with a simulated envelope, the worm plot, the PIT histogram, the residual density against the standard normal, the residuals against the first parameter’s linear predictor and their standard deviation in bins of the second’s: the first three ask whether the family is right, the last three where it is wrong. sample_weight is the aggregation weight of the residuals underneath; engine selects matplotlib or plotly, and only the engine asked for is imported.