2222class SplitConformalRegressor :
2323 """
2424 A conformal regression model using split conformal prediction to generate
25- prediction intervals with statistical guarantees. This method involves
26- using a hold-out conformity set to determine prediction intervals around
27- point predictions from a base regressor.
25+ prediction intervals.
26+
27+ This method involves using a hold-out conformity set to determine
28+ prediction intervals around point predictions from a base regressor.
2829
2930 Parameters
3031 ----------
@@ -243,10 +244,11 @@ def predict(
243244class CrossConformalRegressor :
244245 """
245246 A conformal regression model using cross-conformal prediction to generate
246- prediction intervals with statistical guarantees. This method involves
247- computing conformity scoring across multiple folds in a cross-validation
248- fashion to determine prediction intervals around point predictions from a
249- base regressor.
247+ prediction intervals.
248+
249+ This method involves computing conformity scoring across multiple folds in
250+ a cross-validation fashion to determine prediction intervals around point
251+ predictions from a base regressor.
250252
251253 Parameters
252254 ----------
@@ -272,20 +274,20 @@ class CrossConformalRegressor:
272274 The method used to compute prediction intervals. Options are:
273275 - "base": Based on the conformity scores from each fold.
274276 - "plus": Based on the conformity scores from each fold and
275- the test set predictions.
277+ the test set predictions.
276278 - "minmax": Based on the conformity scores from each fold and
277- the test set predictions, using the minimum and maximum among
278- each fold models.
279+ the test set predictions, using the minimum and maximum among
280+ each fold models.
279281
280282 cv : Union[int, BaseCrossValidator], default=5
281283 The cross-validation strategy used to compute confomity scores.
282284 Valid options:
283285 - integer, to specify the number of folds
284286 - any ``sklearn.model_selection.BaseCrossValidator`` suitable for
285- regression, or a custom cross-validator inheriting from it.
286- Main variants in the cross conformal setting are:
287- - ``sklearn.model_selection.KFold`` (vanilla cross conformal)
288- - ``sklearn.model_selection.LeaveOneOut`` (jackknife)
287+ regression, or a custom cross-validator inheriting from it.
288+ Main variants in the cross conformal setting are:
289+ * ``sklearn.model_selection.KFold`` (vanilla cross conformal)
290+ * ``sklearn.model_selection.LeaveOneOut`` (jackknife)
289291
290292 n_jobs : Optional[int], default=None
291293 The number of jobs to run in parallel when applicable.
@@ -304,7 +306,7 @@ class CrossConformalRegressor:
304306 An array containing the prediction intervals with shape:
305307 - `(n_samples, 2)` if `confidence_level` is a single float
306308 - `(n_samples, 2, n_confidence_levels)` if `confidence_level`
307- is a list of floats.
309+ is a list of floats.
308310
309311 Examples
310312 --------
@@ -492,7 +494,7 @@ def predict(
492494 Generates point predictions for the input data `X`:
493495 - using the model fitted on the entire dataset
494496 - or if aggregation_method is provided, aggregating predictions from
495- the models fitted on each fold
497+ the models fitted on each fold
496498
497499 Parameters
498500 ----------
@@ -524,9 +526,11 @@ def predict(
524526class JackknifeAfterBootstrapRegressor :
525527 """
526528 A conformal regression model using the jackknife-after-bootstrap approach
527- to generate prediction intervals with statistical guarantees. This method
528- combines bootstrap sampling with the jackknife technique to produce robust
529- prediction intervals around point predictions from a base regressor.
529+ to generate prediction intervals.
530+
531+ This method combines bootstrap sampling with the jackknife technique
532+ to produce robust prediction intervals around point predictions from
533+ a base regressor.
530534
531535 Parameters
532536 ----------
@@ -553,9 +557,9 @@ class JackknifeAfterBootstrapRegressor:
553557 The method used for jackknife-after-bootstrap prediction. Options are:
554558 - "base": Based on the conformity scores from each bootstrap sample.
555559 - "plus": Based on the conformity scores from each bootstrap sample and
556- the testing prediction.
560+ the testing prediction.
557561 - "minmax": Based on the minimum and maximum conformity scores from
558- each bootstrap sample.
562+ each bootstrap sample.
559563
560564 n_bootstraps : int, default=100
561565 The number of bootstrap resamples to generate for the
@@ -718,8 +722,9 @@ def predict(
718722class ConformalizedQuantileRegressor :
719723 """
720724 A conformal quantile regression model that generates prediction intervals
721- with statistical guarantees using quantile regression as the base
722- estimator. This approach provides prediction intervals by leveraging
725+ using quantile regression as the base estimator.
726+
727+ This approach provides prediction intervals by leveraging
723728 quantile predictions and applying conformal adjustments to ensure coverage.
724729
725730 Parameters
@@ -790,10 +795,6 @@ def __init__(
790795 estimator : RegressorMixin = QuantileRegressor (),
791796 confidence_level : Union [float , List [float ]] = 0.9 ,
792797 conformity_score : Union [str , BaseRegressionScore ] = "absolute" ,
793- # n_jobs: Optional[int] = None
794- # Not yet available in MapieQuantileRegressor
795- # verbose: int = 0,
796- # Idem
797798 random_state : Optional [Union [int , np .random .RandomState ]] = None ,
798799 ) -> None :
799800 pass
0 commit comments