0

I'm using the partialPlot() function to build PDP of the classification random forests and coming across a question: Do values on y axis represent log(P/1-P) where P represents the probability of the outcome which is a binomial variable? If so, what should I report about this plot in the Results? Thanks.

Here's part of the codes:

rf.sb = randomForest(y ~ x1 + x13 +x14+ x15 + x2 + x3 + x4 + x5 + x6 + x7 + 
                 + x8, data = newsbrf,mtry = 3)
par(bg = "white")
partialPlot(x=rf.sb, pred.data = newsbrf, x.var=x13,  
        ylab = "logit(P) of polytherapy",xlab = "age(yrs)",
        col= "coral", lwd = 2,which.class = "1",          
        xlim=c(5,50))

y represents the use of medications, "0" represents single medication and "1" represents multiple medications.

Here's the plot:

enter image description here

camille
  • 16,432
  • 18
  • 38
  • 60
Ran
  • 1
  • 1
  • What libraries are you using? – camille May 08 '19 at 17:43
  • @camille randomForest – Ran May 08 '19 at 19:40
  • It's helpful to put information like that in your post, since there are several packages with functions called `randomForest`. [See here](https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example) on making reproducible examples—we don't have access to your data – camille May 08 '19 at 19:48

0 Answers0