On the overestimation of random forest’s out-of-bag error

Background The ensemble method random forests has become a popular classification tool in bioinformatics and related fields. The out-of-bag error is an error estimation technique which is often used to evaluate the accuracy of a random forest as well as for selecting appropriate values for tuning parameters, such as the number of candidate predictors that are randomly drawn for a split, referred to as mtry. However, for binary classification problems with metric predictors it was shown that the out-of-bag error overestimates the true prediction error. Based on simulated and real data this paper aims to identify settings for which the overestimation is likely. Moreover, the overestimation was shown to depend on the parameter mtry. Therefore, it is questionable if the out-of-bag error can be used in classification tasks for selecting tuning parameters like mtry. Results The simulation-based and real-data based studies with metric predictor variables show that the overestimation is largest in balanced settings and in settings with few observations , a large number of predictor variables, small correlations between predictors and weak effects. There was hardly any impact of the overestimation on tuning parameter selection. Conclusions Although the prediction performance of random forests was not substantially affected when using the out-of-bag error for tuning parameter selection in the present studies, one cannot be sure that this applies to all future data. For settings with metric predictor variables it is therefore recommended to always use stratified subsampling for both tuning parameter selection and error estimation in random forests. This yielded less biased estimates of the true prediction error.