Greetings to all,
I hope this message finds you well.
In the context of machine learning, particularly deep learning, I would like to clarify a specific point regarding categorical variables. After encoding categorical features, either through label encoding or one-hot encoding, is it necessary or beneficial to scale these encoded variables before feeding them into the model?
I would greatly appreciate your insights or guidance on this matter.
Thank you in advance for your help!
Best regards.
Hello,
what do you mean by scaling the encoded variables? what will the one-hot vector look like after gets scaled?
I hope I can help.
Kind regards
Hello, and thank you for your response.
For example, using Standard Scaller in Python to scale features, even those features are already encoded: Colum location (Rural, Urban ) => (0,1)
my question is about scaling values (0,1)
you are most welcome,
as far as I know, encoding the features as one-hot and feeding them as zeros and ones will enable you to calculate the categorical crossentropy error for them and your program will not work if the labels are in any other form.
I hope I am understanding your question correctly, and the answer can help.
Kind regards
- Badges
- Science topic