OOB ROC curve

[Carlos Crosetti]

Hi, I am experimenting with random forest and was corious aboit what function is used to integrate and find the AUC?

Is Rattle using its own solution to perform numerical integration or is relying on each individual R package to do so?

Thanks, carlos

[Graham Williams]

Hi Carlos,

Rattle's Log tab should give the answer! All commands that you run through the GUI are reported within the Log tab (so you can actually export the Log tab as an R script to repeat the GUI activities). From the Log Tab after building a random forest:

# The `pROC' package implements various AUC functions.

# Calculate the Area Under the Curve (AUC).

pROC::roc(crs$rf$y, as.numeric(crs$rf$predicted))

# Calculate the AUC Confidence Interval.

pROC::ci.auc(crs$rf$y, as.numeric(crs$rf$predicted))

Also under the Evaluate tab if you choose ROC curve you will also see that ROCR is used to calculate AUC there!

# ROC Curve: requires the ROCR package.

library(ROCR)
[...]
performance(pred, "auc")

General philosophy of Rattle is to use other packages to perform the actions, bringing this together into the one GUI.

Hope that helps.

Regards,
Graham

--
You received this message because you are subscribed to the Google Groups "rattle-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rattle-users+unsubscribe@googlegroups.com.
To post to this group, send email to rattle...@googlegroups.com.
Visit this group at https://groups.google.com/group/rattle-users.
For more options, visit https://groups.google.com/d/optout.