Random forest to identify importance of features?
hi,
Is there any way to use random forest operator to get an importance of features ranking?
it could keep track which of their decision trees and subset of features gives best performance on the testing data (out-of-bag-data) and print them out, I know there is a similar implementation of it in R...
Best Answer
-
Hi,
just have a look at Weight by Tree Importance.
~Martin
0
Answers
-
Hi,
just have a look at Weight by Tree Importance.
~Martin
0 -
ok thanks,
its just really confusing that you have to piece together everything from singular operators in RapidMiner, altough it could be implemented together in the other random forest operator
quite confusing if you don't know all the operators in Rapidminer in the beginnning...
0 -
I agree it would be better if the rf simply also returns the weight vector directly. Not sure why it is not like this
0