Train the model. Give it your training data (features `X` and labels `y`). The model learns from this data and stores what it learned internally. For unsupervised methods, `.fit(X)` takes only features.