WEKA 3.7 - Random Forest not numeric, but in WEKA it is...

KNIME Extensions
1

WEKA 3.7 -

The Random Forest classifier is listed as nominal, not numeric. But in WEKA it is also a numeric classifier. Maybe wrong associated?

2

Hi @TomB80,
Do you want to perform classification on numeric data? Or do you want to perform a regression?
The WEKA integration only supports classification, clustering and association rules, not regressions.
To perform regression with the random forest algorithm use either:

best,
Gabriel

2 Likes
3

@gab1one at least these Weka nodes would allow numerical input:

Here are a few examples in action:

1 Like
4

@mlauber71 Thanks, I stand corrected.
@TomB80, sorry for missinforming you on this.

5

Thanks folks,

I maybe was unprecisely. The output column (for regression) is numeric. Within WEKA it workes fine with random forest (and brings better results than the Knime classifier, aditionally it runs more stable and took less memory when trained via batch).

This implementation says it has to be nominal. In Weka it hasn’t…