This is a library for computing the Silhouette Coefficient, a metric used to evaluate the quality of clustering.
Given:
- an array of points (of any dimension),
- an array that corresponds to the cluster assignments for each point,
- an optional custom distance function
this module returns their silhouette score. It measures how similar a data point is to its own cluster and how dissimilar compared to other clusters, providing insights into the cohesion and separation of clusters.
npm install silhouette-coefficient
const silhouette = require("silhouette-coefficient");
//or
//import silhouette from "silhouette-coefficient";
// Example 1
// 1-dimensional data points using the default euclidean distance
const vectors = [[-3], [3], [5]];
const labels = [0, 3, 3];
console.log(silhouette(vectors, labels));
// Expected output: 0.47222222222222215
// ----------------
// Example 2
// 4-dimensional data points using manhattanDistance function to calculate distance
const vectors2 = [
[-7.153, -7.391, 4.773, 1.982],
[-7.391, -7.112, -1.664, 8.179],
[-2.018, 8.284, -6.531, -6.698],
[4.501, 2.63, -6.688, -7.322],
[-5.644, -5.829, -2.458, 8.832],
];
const labels2 = [1, 1, 0, 2, 1];
function manhattanDistance(a, b) {
let dist = 0;
for (let i = 0; i < a.length; i++) {
dist += Math.abs(a[i] - b[i]);
}
return dist;
}
console.log(silhouette(vectors2, labels2, manhattanDistance));
// Expected output: 0.4242836826891837silhouetteCoefficient(vectors, labels, [distanceFormula])
- vectors: An array of data points, where each data point is represented as an array of coordinates.
-
labels: An array indicating the cluster assignment for each data point in vectors.
Example: If labels = [0, 0, 2, 1, 1], the first two points belong to cluster 0, the third point belongs to cluster 2, and the last two points belong to cluster 1. -
distanceFormula: An optional custom distance formula that calculates the distance between two points. It takes two points as arguments and returns a number which represents the distance.
If not provided, it will use Euclidean distance.
The function returns the silhouette score which ranges from -1 to 1, where a higher value indicates better-defined clusters. It can also return NaN if the input data points contain non-numeric values as coordinates.