O'Reilly logo
  • Rajesh k thinks this is interesting:

clone_clf.fit(X_train_folds, y_train_folds)

From

Cover of Hands-On Machine Learning with Scikit-Learn and TensorFlow

Note

how is this possible? The classifier sgd_clf is already fitted with data. that means it already learned the parameters. Why are we fitting a sample of the data again to the classifier?

got it. Clone only coipies the classifier with parameters that are not fit from data.