TimeSeriesSplitStrategy<T>

Models & Types · AiDotNet.Evaluation.CrossValidation

Time Series Split: expanding window cross-validation that respects temporal order.

Implements: ICrossValidationStrategy<T>

For Beginners

Time series data cannot be shuffled because order matters. Time Series Split uses an expanding training window:

  • Training data always comes BEFORE validation data
  • Simulates real-world forecasting where you predict future from past
  • Avoids data leakage from future information

How It Works

Example with 5 splits: Notice how training data grows with each split (expanding window).

Constructors

Constructor Summary
TimeSeriesSplitStrategy(int numSplits, int? maxTrainSize, int? testSize, int gap) Initializes Time Series Split cross-validation.

Properties

Property Summary
int NumSplits { get; }
string Description { get; }
string Name { get; }

Methods

Method Summary
IEnumerable<ValueTuple<int[], int[]>> Split(int dataSize, ReadOnlySpan<T> labels)