0
$\begingroup$

If encode three categorical variables like "bad", "normal", "good" into 0,1,2, after that can I treat them as numerical values? So can I perform on them MinMaxScaler or RobustScaler? Or, since they are from categorical values, I must leave them like 0,1,2.

Thanks for your reply.

$\endgroup$

1 Answer 1

1
$\begingroup$

If there is only a few categorical features and therefore, every feature in the ordinal encoding is close to 0 (in your example: 0,1,2), there is usually no need to perform scaling on these features and even if you scaled them, there should be no visible difference in results.

But theoretically, you can even scale encoded categorical features for training since they are treated by the model as just another numeric input.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.