用什么来分类R中的提升决策树？

Question

I'm rather new to R. I have found ways to run classification using trees with bagging (randomforest), and ways to run regression with boosting (gbm), but what I'm looking for is to do classification using boosting.

Supposedly, gbm is able to do this when the target variable consists of 0s and 1s and when you set Bernoulli as the function -- but no, the results that I get still point to regression being used.

Any help will be very appreciated.

Answer 1

如果您在插入符包中使用xgboost，则可能还必须同时更改objective = binary:logistic和eval_metric =以反映您的分类需求。

Answer 2

It becomes classification if you transform the predicted values to 0 and 1. For example with round(). That is, using a threshold of 0.5. But there are more sophisticated approaches to finding the optimal threshold.