Unsupervised Learning
Finding structure in unlabeled data — clustering, dimensionality reduction — with no correct answers to learn from.
Unsupervised learning finds patterns in data that has no labels — grouping similar points (clustering), compressing dimensions (PCA), or estimating the data’s distribution. There is no “right answer” to predict; the goal is to discover structure.
Worked example: given customer purchase histories with no labels, k-means clustering discovers natural segments (bargain hunters, loyalists, one-time buyers) you did not define in advance, which you can then act on. Gotcha: without labels there is no objective score for “correct,” so evaluating unsupervised results is inherently fuzzy (is this the right number of clusters?) and needs human judgment or downstream task performance — which is why unsupervised methods are often a step within a larger pipeline rather than the final answer.