RollingOriginSplitter<T>
Models & Types · AiDotNet.Preprocessing.DataPreparation.Splitting.TimeSeries
Rolling origin evaluation for multi-step forecasting.
Inheritance: DataSplitterBase<T> → RollingOriginSplitter<T>
For Beginners
Rolling origin is designed for evaluating forecasts at multiple horizons (1-step, 2-step, etc. ahead).
How It Works
How It Works: The “origin” is the last training point, and you forecast H steps ahead.
When to Use:
- Multi-step forecasting evaluation
- When you need to test different forecast horizons
- Standard time series forecasting evaluation
Constructors
| Constructor | Summary |
|---|---|
RollingOriginSplitter() |
Initializes a new instance with default settings. |
RollingOriginSplitter(int initialOrigin, int horizon, int step) |
Creates a new rolling origin 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) |