What does it mean if my IncNodePurity measures are all 0

Question

I ran an initial Random Forest Model on my data in R and tried to calculate the feature importance statistics. For some reason, my IncNodePurity measures were all 0. Any idea why this could happen or what it is I did wrong? Thanks

rf<- randomForest(target~.,
                    data=train,importance=TRUE,proximity=TRUE)
round(importance(rf),2)

Results:

                    %IncMSE IncNodePurity
age                  1.82   0.00
income               6.89   0.00
education           -0.12   0.00
car_ownership        0.99   0.00
health_metric       10.34   0.00

Thanks!

Answer 1

Try using more digits when reporting variable importance. In my models, IncNodePurity is commonly below 0.01. If you are limiting yourself to 2 digits, these values would show as 0.00.