Unsupervised Learning

Finding structure in unlabelled data, with no target to predict and no single right answer to optimise against.

When to use

Recognise it by the lack of labels. Typical tasks are clustering (K-Means) and dimensionality reduction (PCA), evaluated more subjectively with the Elbow Method and Silhouette Score rather than accuracy. The lecture's case is StyleNation customer segmentation. Appears in PS 03 and PS 05.

See DS Lec 5.

Appears in