CombinatorialPurgedSplitter<T>

Models & Types ยท AiDotNet.Preprocessing.DataPreparation.Splitting.TimeSeries

Combinatorial Purged Cross-Validation splitter for time series data.

Inheritance: DataSplitterBase<T> โ†’ CombinatorialPurgedSplitter<T>

For Beginners

This is an advanced cross-validation method designed specifically for financial time series where data leakage can invalidate backtests.

How It Works

How It Works:

  1. Divide data into n groups based on time periods
  2. Generate all combinations of k groups for testing
  3. For each combination, remaining groups form training set
  4. Apply purging (remove samples near test boundaries) and embargo (block after test)

Use Cases:

  • Financial backtesting where overlapping data causes leakage
  • Time series with autocorrelation that persists across samples
  • Generating many independent train/test splits from limited time series

Constructors

Constructor Summary
CombinatorialPurgedSplitter(int nGroups, int nTestGroups, int purgeSize, int embargoSize, int randomSeed) Creates a new Combinatorial Purged CV splitter.

Properties

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

Methods

Method Summary
DataSplitResult<T> Split(Matrix<T> X, Vector<T> y)
IEnumerable<DataSplitResult<T>> GetSplits(Matrix<T> X, Vector<T> y)