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:

  1. Compute empirical CDF for each feature
  2. Calculate tail probabilities for each point
  3. Combine probabilities across features
  4. 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)