PrequentialSplitter<T>
Models & Types · AiDotNet.Preprocessing.DataPreparation.Splitting.Online
Prequential (predictive sequential) evaluation splitter for online/streaming data.
Inheritance: DataSplitterBase<T> → PrequentialSplitter<T>
For Beginners
Prequential evaluation is the standard way to evaluate models on streaming data. For each new sample: first use it as a test sample (predict its label), then add it to the training set. This simulates real-world continuous learning.
How It Works
How It Works:
- Start with an initial training window
- For each subsequent sample:
- Test: Predict using current model
- Train: Update model with the true label
- Evaluation is “test-then-train” on every sample
When to Use:
- Streaming/online learning scenarios
- Concept drift detection
- Real-time prediction systems
- Evaluating adaptive algorithms
Constructors
| Constructor | Summary |
|---|---|
PrequentialSplitter(int initialTrainSize, int batchSize, bool forgetOldSamples, int maxTrainSize, int randomSeed) |
Creates a new prequential evaluation 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) |