Glossary

Unsupervised learning

A machine learning approach that finds structure or patterns in data that has no labeled correct answer to learn from.

Unsupervised learning is machine learning applied to data that has no labeled correct answer: there is no target column to predict, only the input data itself. Instead of learning a mapping from inputs to a known output, as in supervised learning, an unsupervised method looks for structure, groupings or patterns that already exist in the data.

The two most common tasks are clustering, grouping similar records together, for example with k-means clustering, and dimensionality reduction, compressing many correlated variables down to a smaller set that still captures most of the information. A related application is anomaly detection, flagging records that do not fit the patterns the rest of the data shows.

Unsupervised learning matters where labels do not exist or would be too expensive to collect, and as an exploratory step before a supervised project, to understand a dataset's natural structure before deciding what to predict and how. Because there is no ground truth to check predictions against, evaluating results is inherently more subjective than in supervised learning, and results depend heavily on choices such as the similarity measure or number of clusters, which is why unsupervised output is usually paired with domain review rather than treated as a final answer on its own.

Last reviewed September 22, 2026

In the index now

Related terms

Supervised learning A machine learning approach where a model learns from examples that already include the correct answer, to predict that answer on new data. Clustering An unsupervised technique that groups data points so items within a group are more similar to each other than to items outside it. Dimensionality reduction Techniques that compress a dataset with many variables into fewer dimensions while preserving as much meaningful information as possible. Anomaly detection Identifying data points or patterns that deviate significantly from expected or normal behavior. K-means clustering A clustering algorithm that partitions data into k groups by repeatedly assigning points to the nearest of k cluster centers.