Dear R-User!
This is a question to all of you that are familiar with the 'dismo'
package, in particular we are interested in the gbm.step() function to
create a boosted regression tree model in R. From the model output object we
can use the $cv.statistics to get information on the cross-validation model
performance (e.g. cross-validation AUC). However, quite often the AUC has
been criticized for various reasons (e.g. Lobo et al. 2008) and it was
proposed also to use other measures of model performance (e.g. sensitivity
and specificity, Cohen's kappa, TSS).
Thus I am wondering wondering which of the $cv.statistical output is the
most appropriate indicator of model performance (e.g. for comparing two
boosted regression tree models). Moreover I'd like to know if it is somehow
possible to extract/calculate values of specificity and sensitivity from
the output model object (i.e without re-calculating the model). This would
be valuable information to calculate e.g. a kappa value or TSS value from a
gbm.step - model-object.
Thanks for your answers,
Best,
Johannes
[[alternative HTML version deleted]]
______________________________________________
[email protected] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
