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

Key points

• Data leakage occurs when test information influences training.
• Scaling parameters must be derived solely from training data.
• Test data must be treated as unseen during the fitting phase.