ix-cluster

name	ix-cluster
description	Cluster data using K-Means or DBSCAN
disable-model-invocation	true

Cluster

Group data points into clusters using unsupervised learning.

When to Use

When the user has unlabeled data and wants to find natural groupings, segments, or patterns.

Algorithms

• K-Means — Fast, works well when clusters are spherical and k is known
• DBSCAN — Density-based, finds arbitrary-shaped clusters, handles noise/outliers

Choosing k

If the user doesn't specify k, suggest the elbow method:

1. Run K-Means for k=2..10
2. Plot inertia (sum of squared distances to centroids)
3. Pick the "elbow" where improvement plateaus

Programmatic Usage

use ix_unsupervised::kmeans::KMeans;
use ix_unsupervised::dbscan::DBSCAN;

// K-Means
let model = KMeans::new(k, max_iter, seed);
let assignments = model.fit(&data);

// DBSCAN
let model = DBSCAN::new(eps, min_points);
let labels = model.fit(&data); // -1 = noise

Interpretation

• Report cluster sizes and centroids
• Flag if any cluster is disproportionately large (may need more k)
• For DBSCAN, report number of noise points