[R] Which column in randomForest importances (for regression) is MSE and which IncNodePurity

Dimitri Liakhovitski ld7631 at gmail.com
Wed May 5 19:04:37 CEST 2010

I've run the function randomForest with importance=T. All my variables
(predictors and the dependent variable) are numeric.
rf<-randomForest(formula, data=mydata, importance=T, etc.)

my results object "rf" contains predictor importances:

I am seeing two columns:

      %IncMSE IncNodePurity
V1 -0.01683558      58.10910
V2  0.04000299      71.27579
V3  0.01974636      67.22586
V4  0.25020393     113.69823
V5  0.03146358      67.11151
V6  0.01717313      66.57246
V7 -0.00500985      62.37103
V8 -0.02862065      66.15369
V9 -0.02431507      54.50013

They seem to be clearly labeled %IncMSE and IncNodePurity

However, when I look in ?randomForest, I am reading about importance as a
component of my rf object:
A matrix with nclass + 2 (for classification) or two (for regression)
columns. For classification, the first nclass columns are the class-specific
measures computed as mean descrease in accuracy. The nclass + 1st column is
the mean descrease in accuracy over all classes. The last column is the mean
decrease in Gini index. *For Regression, the first column is the mean
decrease in accuracy and the second the mean decrease in MSE. If
importance=FALSE, the last measure is still returned as a vector.*

Maybe I am confused for no reason - but which column is which?
Is %IncMSE = mean decrease in accuracy?

Thanks a lot for clarifying!

Dimitri Liakhovitski
Dimitri.Liakhovitski at ninah.com

	[[alternative HTML version deleted]]

R-help at r-project.org mailing list
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

More information about the R-help mailing list