Why is it critical to fit a scaler only on the training set before transforming the test set?

Question

Data Science — Hard

Why is it critical to fit a scaler only on the training set before transforming the test set?

Accepted Answer

Fitting a scaler on the entire dataset (including the test set) allows information from the test set to 'leak' into the training process. This leads to overly optimistic performance estimates because the model effectively 'sees' the distribution of the test data during training. The scaler must be fit on training data and then applied to the test data to maintain a strict separation between training and evaluation.