Loading…
Glossary term
Learning patterns from data that has no labels, such as grouping similar items.
Unsupervised learning looks for structure in data that has no labels. It can group similar items together (clustering), compress data to its main directions, or spot unusual points.
Because no answers are provided, results need human interpretation to decide whether they are useful.
Find groups with no labels
from sklearn.cluster import KMeans
groups = KMeans(n_clusters=3, n_init=10).fit_predict(X)Most data in the world is unlabelled, so methods that can learn from it directly are valuable for exploration and for pretraining.