I am applying logistic regression and I am using a mixture of continuous and binary features. My question would be how do you go about binary features when trying to normalize. I would very much appreciate any help.
Asked
Active
Viewed 825 times
6
-
I don't think you really need to do any normalization. Clearly your means and variances are already in the same magnitudes. – Thomas Ahle Mar 18 '14 at 11:59
-
Why do you want to normalize your data in the first place? – Alejandro Ochoa Jul 02 '15 at 01:45
1 Answers
2
Normalization means that you put your data in a particular range, often $0$ to $1$. If you have coded your binary variable with $0$ and $1$, you already have this property and do not need to do anything. If you use a different type of coding of your binary variable, such as $\pm 1$, then the usual $\dfrac{x_i - \min(x)}{\max(x) - \min(x)}$ should work.
If you have not coded your categories with numbers, your software does it under the hood, and the documentation should say how.
Dave
- 62,186