metacausal.aggregation.FoldSpec

class metacausal.aggregation.FoldSpec(train_indices, test_indices, n_folds)

Bases: object

Result of a data split: absolute indices into the original array per fold.

Variables:

• train_indices (list of arrays) – train_indices[j] contains the row indices used for training in fold j.

• test_indices (list of arrays) – test_indices[j] contains the row indices held out for evaluation in fold j. For CrossFitSplit these are out-of-fold (OOF) indices. For TrainAvgSplit this is the “averaging set” where weights are optimized.

• n_folds (int) – Number of folds. Equals len(train_indices) == len(test_indices).

Parameters:

• train_indices (list[ndarray])

• test_indices (list[ndarray])

• n_folds (int)

Methods

__init__

Attributes

train_indices
test_indices
n_folds

Details

n_folds: int

test_indices: list[ndarray]

train_indices: list[ndarray]