← Quiz: Evaluation and Tuning: all 10 questions
Quiz: Evaluation and Tuning · Question 9 of 10
Why is scaling the whole dataset before splitting it a form of data leakage?
Why is scaling the whole dataset before splitting it a form of data leakage?
- A.Scaling changes the labels
- B.It deletes rows
- C.It makes training slower
- D.The scaler learns the mean and spread from test rows, so test information leaks into training
Answer & explanation
Correct answer: The scaler learns the mean and spread from test rows, so test information leaks into training
Fit preprocessing on training data only; a Pipeline does this for you.
Can you get the rest right without peeking?
Take the timed Quiz: Evaluation and Tuning quiz free. Pass it to earn XP and count towards your certificate.
Start the quiz →