LSCPDetector<T>
Models & Types · AiDotNet.AnomalyDetection.Ensemble
Detects anomalies using LSCP (Locally Selective Combination in Parallel Outlier Ensembles).
Inheritance: ModelBase<T, Matrix<T>, Vector<T>> → AnomalyDetectorBase<T> → LSCPDetector<T>
Implements: IGeneratedParameterRegistrar<T>
For Beginners
LSCP improves ensemble detection by selecting the most competent detectors for each test point locally. Instead of combining all detectors equally, it selects the best performing detectors based on local pseudo ground truth.
How It Works
The algorithm works by:
- Train multiple diverse base detectors
- For each test point, find local region
- Evaluate detector competence in that region
- Combine scores from most competent detectors
When to use:
- When different detectors work better in different regions
- For heterogeneous anomaly types
- When ensemble averaging isn’t optimal
Industry Standard Defaults:
- N estimators: 10
- Local region size: 30 neighbors
- Contamination: 0.1 (10%)
Reference: Zhao, Y., et al. (2019). “LSCP: Locally Selective Combination in Parallel Outlier Ensembles.” SDM.
Constructors
| Constructor | Summary |
|---|---|
LSCPDetector(int nEstimators, int localRegionSize, double contamination, int randomSeed) |
Creates a new LSCP anomaly detector. |
Properties
| Property | Summary |
|---|---|
int LocalRegionSize { get; } |
Gets the local region size. |
int NEstimators { get; } |
Gets the number of estimators. |
Methods
| Method | Summary |
|---|---|
Vector<T> ScoreAnomalies(Matrix<T> X) |
|
void Fit(Matrix<T> X) |