XGBODDetector<T>
Models & Types · AiDotNet.AnomalyDetection.Ensemble
Detects anomalies using XGBOD (Extreme Gradient Boosting Outlier Detection).
Inheritance: ModelBase<T, Matrix<T>, Vector<T>> → AnomalyDetectorBase<T> → XGBODDetector<T>
Implements: IGeneratedParameterRegistrar<T>
For Beginners
XGBOD is a semi-supervised method that creates new features from unsupervised outlier detection scores, then trains a supervised classifier on these enhanced features. It combines the best of both worlds.
How It Works
The algorithm works by:
- Train multiple unsupervised outlier detectors
- Generate outlier scores as new features (TOS: Transformed Outlier Scores)
- Combine TOS with original features
- Train gradient boosting classifier on enhanced features
When to use:
- When you have some labeled anomaly examples
- When feature engineering with outlier scores helps
- As a powerful ensemble method
Industry Standard Defaults:
- N estimators: 10 unsupervised detectors
- Boosting rounds: 100
- Contamination: 0.1 (10%)
Reference: Zhao, Y., Hryniewicki, M.K. (2018). “XGBOD: Improving Supervised Outlier Detection with Unsupervised Representation Learning.” IJCNN.
Constructors
| Constructor | Summary |
|---|---|
XGBODDetector(int nEstimators, int boostingRounds, double contamination, int randomSeed) |
Creates a new XGBOD anomaly detector. |
Properties
| Property | Summary |
|---|---|
int BoostingRounds { get; } |
Gets the number of boosting rounds. |
int NEstimators { get; } |
Gets the number of base estimators. |
Methods
| Method | Summary |
|---|---|
Vector<T> ScoreAnomalies(Matrix<T> X) |
|
void Fit(Matrix<T> X) |