Random Forest poor performance

Jun 26, 2015 at 9:21 AM
Hi guys,
I'm trying to make the the typical prediction on customers promotion response and I'd like to use the RandomForest script. I have some storical data in a report in the format:

ID G1 G2 G3 Age R F M OP TARGET
49347 1 0 0 1 1 1 3 1 0
76103 1 0 0 2 2 3 1 1 1
74063 1 0 0 2 1 1 1 1 0
141992 1 0 0 3 1 1 2 0 0
142209 1 0 0 3 3 4 1 0 0
143713 1 0 0 3 3 3 1 0 0
145063 1 0 0 3 1 2 4 1 0
159296 1 0 0 2 1 2 1 1 0
160773 1 0 0 3 1 1 1 1 1

and the R metric for training is:
RScript<[BooleanParam9]=True,[NumericParam1]="750",[NumericParam2]="3",[NumericParam3]="42",[StringParam9]="RandomForest",[_OutputVar]="ClassId",[_RScriptFile]="RandomForest.R">([TARGET],G1, G2, G3, Age,R,F,M,OP)

I left default values for the params. Everything works fine but, the prediction values got in this training step are so strange, I get only 4 or 5 positive value among 10000 records. Should I try with different values of params? f.i. the seed number or the decision trees numbers?
Did anyone try to use the script?

thank you.
Jul 2, 2015 at 6:01 AM
Ok got it. The script is ok but my data are strongly unbalanced.
Marked as answer by mortommy on 7/1/2015 at 10:03 PM