AnchoredWalkForwardSplitter<T>
Models & Types ยท AiDotNet.Preprocessing.DataPreparation.Splitting.TimeSeries
Anchored walk-forward validation with a fixed starting point.
Inheritance: DataSplitterBase<T> โ AnchoredWalkForwardSplitter<T>
For Beginners
This is similar to regular walk-forward, but the training always starts from the same point (anchored) rather than sliding.
How It Works
How It Works: Training always starts at index 0 (the anchor).
When to Use:
- When historical data remains relevant
- When you want maximum training data
- Traditional time series forecasting
Constructors
| Constructor | Summary |
|---|---|
AnchoredWalkForwardSplitter() |
Initializes a new instance with default settings. |
AnchoredWalkForwardSplitter(int initialTrainSize, int testSize, int? stepSize) |
Creates a new anchored walk-forward 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) |