For supervised machine learning for prediction, if I had some feature variables that are real, and also some features that are categorical--which have been coded using dummy variables (010, 001 etc)--I have normalized the real variables so that each of them sums to one. I am wondering what kind of a preprocessing I should do for the rest of the categorical features, before I run cross-validation routines and regression methods.
$\begingroup$
$\endgroup$
2
-
1$\begingroup$ Why have you normalised the real ones to add to 1? I don't think you need to normalise the dummy ones. $\endgroup$– Dirk NCommented Sep 18, 2012 at 21:38
-
$\begingroup$ I need a more concrete answer than an "I think" $\endgroup$– qlinckCommented Sep 18, 2012 at 22:19
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Normalizing categorical variables ought not have any real effect.
answered Sep 30, 2012 at 22:18