Unsupervised Learning

Unsupervised learning finds structure in data when no target label tells you what the answer should be. You start with observations and their features. The method returns a description of patterns it can find, not a prediction proven correct by known answers.

That distinction changes the work. In supervised learning, a label gives you a direct score. In unsupervised learning, you must decide whether a pattern is useful, stable, and meaningful in its real context.

What it is for

Use unsupervised learning to explore a dataset before you know its important segments or factors. Common goals are: