Random Ensemble Decision Trees for Learning Concept-Drifting Data Streams

Abstract

Few online classification algorithms based on traditional inductive ensembling focus on handling concept drifting data streams while performing well on noisy data. Motivated by this, an incremental algorithm based on random Ensemble Decision Trees for Concept-drifting data streams (EDTC) is proposed in this paper. Three variants of random feature selection are developed to implement split-tests. To better track concept drifts in data streams with noisy data, an improved two-threshold-based drifting detection mechanism is introduced. Extensive studies demonstrate that our algorithm performs very well compared to several known online algorithms based on single models and ensemble models. A conclusion is hence drawn that multiple solutions are provided for learning from concept drifting data streams with noise.