Interpreting negative value in varImp () random forest

Question

Interpreting negative value in varImp () random forest

I am using a function randomForest()

from the randomForest package. I am trying to understand the output given by the varImp()

caret package function . It shows negative values for several variables. What does a negative value indicate?

Below is the result:

           Overall
Year        6.7712260
Month      -2.6684553
DOW         2.2681865
City.Group -0.3534247
Type        1.4049568
P1          8.5207072
P2          6.0449333
P3          5.3579060
P4          4.1441309
P5          5.9207454
P6          7.3744485
P7          3.6858546

Thanks in advance!

+3

r machine-learning random-forest

Arushi Apr 19 15 at 15:19

source to share

No one has answered this question yet

See similar questions:

8

What does negative% IncMSE mean in the RandomForest package?

or similar:

629

data.table vs dplyr: can something do good and other do bad or bad?

3

extract predictor names from varImp

2

How do I get a VarImp plot for glmnet to display variable severity in both directions?

1

Using varImp returns Error: $ operator is not valid for atomic vectors

1

Difference between importance value (random forest) and RandomForest $ value

1

Why do different random implementations of a forest in R give different results?

1

random forest with set false positive and sensitive

0

varImp adds unusual suffixes to variable names

0

Random forest with R:% IncMSE mismatch between varImp (model) and model $ importance [, 1]

0

Problems of creating a graph for varImp

Interpreting negative value in varImp () random forest

More articles: