|
Hi
I successfully execute a DRF (using the iris data) with ALGLIB, the only way that the predicted nominal variable could be encoding was using (as the "General principles" [http://www.alglib.net/dataanalysis/generalprinciples.php#header3] suggests) 0, 1, 2. And everything worked smoothly.
Now, I want to use the just made code, but the predictors will include nominal variables, so my question is How the code will recognize wich variables are nominal and which aren' t, if I use the integer encoding? I must put the numerical variables as real ones and the nominal as integers? That's it?
If that approach doesn't work, I must use (following again the "General principles" page) the "1-of-N" encoding, so if my nominal variable has three possible values I have to substitute that column for three columns?
Thank you for your patience
Adolfo
|
Login
Register
FAQ
Search
