CovariateShiftSplitter<T>
Models & Types ยท AiDotNet.Preprocessing.DataPreparation.Splitting.Specialized
Covariate shift splitter that creates intentional distribution shift between train and test.
Inheritance: DataSplitterBase<T> โ CovariateShiftSplitter<T>
For Beginners
Covariate shift occurs when the distribution of input features differs between training and test data, even though the relationship between inputs and outputs remains the same. This splitter intentionally creates such a shift.
How It Works
How It Works:
- Identify a primary feature dimension for the shift
- Select training samples from one range of that feature
- Select test samples from a different range
When to Use:
- Testing model robustness to distribution shift
- Simulating real-world deployment scenarios
- Evaluating domain adaptation techniques
Constructors
| Constructor | Summary |
|---|---|
CovariateShiftSplitter(double testSize, double shiftStrength, int shiftDimension, int randomSeed) |
Creates a new covariate shift splitter. |
Properties
| Property | Summary |
|---|---|
string Description { get; } |
Methods
| Method | Summary |
|---|---|
DataSplitResult<T> Split(Matrix<T> X, Vector<T> y) |