Clusters.kmeansPlusPlus

kmeansPlusPlus(
    X: number[][],
    n_clusters: number,
    sampleWeight?: number[],
    randomState: () => number = Math.random
): { centers: number[][]; indices: number[] }

This utility initializes cluster centers using the k-means++ strategy.

const { centers } = kmeansPlusPlus(X, 3);

Practical guide: kmeansPlusPlus in JavaScript and TypeScript

kmeansPlusPlus provides improved centroid initialization to make KMeans more stable and accurate.

When to use kmeansPlusPlus

• Random initialization creates unstable cluster assignments.
• You need faster convergence and better centroid quality.
• You run repeated clustering jobs in production pipelines.

Implementation workflow

1. Initialize centroids with k-means++ strategy before training.
2. Run KMeans fitting with the initialized seeds and compare inertia.
3. Retain the best run by objective score and validation diagnostics.

JavaScript deployment notes

• Prefer feature scaling for distance-based and gradient-based algorithms to improve stability.
• In browser apps, run heavy training in Web Workers to keep UI interactions smooth.
• Keep a simple baseline from the same module as a fallback model for comparison.

Search intents this page targets

• kmeansPlusPlus JavaScript
• kmeansPlusPlus TypeScript
• kmeansPlusPlus browser machine learning
• @kanaries/ml kmeansPlusPlus