ICrossValidationStrategy<T>

Interfaces · AiDotNet.Evaluation.CrossValidation

Defines a cross-validation splitting strategy.

Known implementations: BlockedKFoldStrategy<T>, BootstrapStrategy<T>, GroupKFoldStrategy<T>, KFoldStrategy<T>, LeaveOneOutStrategy<T>, LeavePOutStrategy<T>, MonteCarloStrategy<T>, NestedCVStrategy<T>, PurgedKFoldStrategy<T>, RepeatedKFoldStrategy<T>, ShuffleSplitStrategy<T>, SlidingWindowStrategy<T>, StratifiedGroupKFoldStrategy<T>, StratifiedKFoldStrategy<T>, TimeSeriesSplitStrategy<T>

For Beginners

Cross-validation strategies determine how to split your data into training and validation sets. Different strategies are appropriate for different data types:

  • K-Fold: Standard approach for most data
  • Stratified K-Fold: Preserves class distribution (for classification)
  • Time Series Split: Respects temporal order (for time series)
  • Leave-One-Out: Maximum data usage but computationally expensive

Properties

Property Summary
int NumSplits { get; } Gets the number of splits (folds) this strategy will generate.
string Description { get; } Gets a description of this strategy suitable for documentation.
string Name { get; } Gets the name of this cross-validation strategy.

Methods

Method Summary
IEnumerable<ValueTuple<int[], int[]>> Split(int dataSize, ReadOnlySpan<T> labels) Generates train/validation index splits for the given data size.