Clustering

Clustering is an unsupervised machine learning technique that organises data points into groups so that points in the same group are more similar to each other than to those in other groups. There are no target labels; the structure comes from a distance or similarity measure. Common algorithms include k-means, hierarchical clustering and DBSCAN.

Why it matters

Clustering helps discover customer segments, group similar documents, spot unusual behaviour and compress large datasets into representative groups. Because nothing tells the algorithm what is "correct", results depend on the features, scaling and the chosen number of clusters, and you must judge them for usefulness as much as for statistical scores.

Example

Customers described by (orders per month, average basket value)
k-means with k = 3 might yield:
  Cluster A: few orders, small baskets
  Cluster B: many orders, small baskets
  Cluster C: few orders, large baskets

Marketing then tests different messages for each group.

Common mistakes

• Not scaling features, so one large-valued column dominates the distance.
• Choosing k arbitrarily without checking stability or interpretability.
• Treating clusters as ground truth when they may just reflect noise.

Related

• Unsupervised learning
• Classification
• Outlier
• Data science

This page is general information written by our editorial team and updated when documentation changes. Servers and providers differ, so test changes safely first. See our editorial policy and disclaimer.