Hope you are doing well. I have some problem with mixed. For the classification problem, can we standardize categorical and numerical variables together or just standardise numerical variable or don't need to standardise. For example age[0..120], sex[0..2] , productName[0..5], etc..
$\begingroup$
$\endgroup$
1
-
$\begingroup$ Just do it for Numerical features. $\endgroup$– 10xAICommented Mar 31, 2021 at 11:25
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
Each feature is standardized independently so it does not matter if different features are different types (numeric or categorical).
Standardized is typically only done on numerical features.
answered Jun 27, 2021 at 15:26
$\begingroup$
$\endgroup$
It is better to apply for all features individually.
Besides of that, if categorical variables do not have so many discrete values, you can consider one hot encoding as well. This approach generally get better results.
