水随机森林的置信区间

Question

I noticed that there are a couple functions designed to calculate the confidence interval for models built using randomForest packages, such as rfPredVar in RFinfer . I am wondering if anyone knows any functions like rfPredVar that can calculate the confidence interval for RF models built using h2o package. Many thanks in advance.

Answer 1

At this point, there is no built-in functionality for generating confidence intervals within H2O.

As an aside: It says in the docs that the RFInfer package only works for regression trees. So if you happen to be performing binary classification instead of regression and want CIs for the AUC or cross-validated AUC estimate of model performance, then you can use a package I wrote called cvAUC for your Random Forest predictions or any type of model predictions.