In the C3_W2_RecSysNN_Assignment, the data was normalized before splitting into training and testing sets. Isn’t this a form of data leakage? Shouldn’t normalization be done after splitting, using only the training data?
1 Like
Hello, @maab,
Welcome to this community!
Yes, I am with you on this that those scalars should be fitted with training data only, as the testing data is not supposed to be known at this stage. I will share your post with the course team for their review.
Cheers,
Raymond
1 Like
Thanks for the response!
1 Like