ConformalPredictor

A calibrated conformal predictor ready to produce intervals.

Usage

Source

ConformalPredictor(
    model,
    calibration_scores,
    quantile,
    level,
    method,
    _models=None,
    _fold_preds=None,
    _fold_ids=None,
)

Created by conformal_fit(). Wraps a fitted GAM (or, for "cv+"/"jackknife+", an ensemble of fold/leave-one-out GAMs) together with the conformity scores from calibration, so that predict() on new data returns intervals with a finite-sample marginal coverage guarantee that does not rely on the GAM’s error distribution being correctly specified — only on the calibration and test data being exchangeable.

Attributes

model: GAM

The GAM used for point predictions: fit on the training split for "split", or on the full data for "cv+"/"jackknife+" (in which case predictions are instead ensembled from the per-fold/per-observation models).

calibration_scores: NDArray

Absolute residuals from the calibration step (calibration split, K-fold, or leave-one-out, depending on method).

quantile: float

The calibration quantile of calibration_scores used to set interval half-width ("split" only).

level: float

Nominal coverage level.

method: str

Conformal method: "split", "cv+", or "jackknife+".

Methods

Name	Description
predict()	Produce conformal prediction intervals on new data.

---

predict()

Produce conformal prediction intervals on new data.

Usage

Source

predict(new_data)

Dispatches to the interval construction appropriate for self.method: simple values +/- quantile bands for "split", or the min/max-based "cv+" and "jackknife+" constructions that combine every fold/leave-one-out model’s prediction with every calibration residual.

Parameters

new_data: InputData

Column-oriented covariate data.

Returns

ConformalResult