ECODDetector<T>
Models & Types · AiDotNet.AnomalyDetection.Probabilistic
Detects anomalies using ECOD (Empirical Cumulative Distribution Functions for Outlier Detection).
Inheritance: ModelBase<T, Matrix<T>, Vector<T>> → AnomalyDetectorBase<T> → ECODDetector<T>
Implements: IGeneratedParameterRegistrar<T>
For Beginners
ECOD uses the cumulative distribution function (CDF) to identify outliers. Points with extreme values in any dimension have low probability under the empirical distribution and are flagged as anomalies.
How It Works
The algorithm works by:
- Compute empirical CDF for each feature
- Calculate tail probabilities for each point
- Combine probabilities across features
- Points with very low probabilities are anomalies
When to use:
- Large datasets (linear complexity O(n))
- High-dimensional data
- When you need a fast, parameter-free method
Industry Standard Defaults:
- No parameters to tune (parameter-free)
- Contamination: 0.1 (10%)
Reference: Li, Z., et al. (2022). “ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functions.” IEEE TKDE.
Constructors
| Constructor | Summary |
|---|---|
ECODDetector(double contamination, int randomSeed) |
Creates a new ECOD anomaly detector. |
Methods
| Method | Summary |
|---|---|
Vector<T> ScoreAnomalies(Matrix<T> X) |
|
void Fit(Matrix<T> X) |