WCSS<T>
Models & Types · AiDotNet.Clustering.Evaluation
Within-Cluster Sum of Squares (WCSS) metric for evaluating cluster compactness.
Implements: IClusterMetric<T>
For Beginners
WCSS measures how “tight” your clusters are.
For each point:
- Find the center of its cluster
- Measure the distance squared
- Add up all these distances
Lower WCSS = Better clustering (tighter clusters)
Also called “inertia” in scikit-learn.
How It Works
WCSS measures the total squared distance of each point to its cluster centroid. Lower values indicate more compact clusters.
WCSS = sum over all clusters k of sum over all points i in k of ||x_i - c_k||^2
Constructors
| Constructor | Summary |
|---|---|
WCSS() |
Initializes a new WCSS instance. |
Properties
| Property | Summary |
|---|---|
string Name { get; } |
Methods
| Method | Summary |
|---|---|
Dictionary<int, T> ComputePerCluster(Matrix<T> data, Vector<T> labels) |
Computes WCSS per cluster. |
double Compute(Matrix<T> data, Vector<T> labels) |