KMeansDetector<T>
Models & Types · AiDotNet.AnomalyDetection.ClusterBased
Detects anomalies using K-Means clustering distance.
Inheritance: ModelBase<T, Matrix<T>, Vector<T>> → AnomalyDetectorBase<T> → KMeansDetector<T>
Implements: IGeneratedParameterRegistrar<T>
For Beginners
K-Means clustering groups data into k clusters based on distance to cluster centers (centroids). Points far from their nearest centroid are considered anomalies because they don’t fit well into any cluster.
How It Works
The algorithm works by:
- Partition data into k clusters using K-Means
- For each point, find distance to nearest centroid
- Points with large distances are anomalies
When to use:
- When data naturally forms spherical clusters
- When you have a rough idea of the number of clusters
- Fast and scalable to large datasets
Industry Standard Defaults:
- K (clusters): 8 (or estimated from data)
- Max iterations: 100
- Contamination: 0.1 (10%)
Constructors
| Constructor | Summary |
|---|---|
KMeansDetector(int k, int maxIterations, double contamination, int randomSeed) |
Creates a new K-Means anomaly detector. |
Properties
| Property | Summary |
|---|---|
int K { get; } |
Gets the number of clusters. |
int MaxIterations { get; } |
Gets the maximum number of iterations. |
Methods
| Method | Summary |
|---|---|
Vector<T> ScoreAnomalies(Matrix<T> X) |
|
void Fit(Matrix<T> X) |