Which preprocessing step is required to prevent data leakage when using cross-validation?

Question

Data Science — Medium

Which preprocessing step is required to prevent data leakage when using cross-validation?

Accepted Answer

Data leakage occurs when information from the test set (or validation fold) influences the training process. If a scaler is fit on the entire dataset, the training data 'sees' the distribution (mean/std) of the test data. To prevent this, the scaler must be fit only on the training portion of each fold and then applied to the validation portion.