Hybrid dynamic chunk ensemble model for multi-class data streams

In the analysis more specifically in classification of continuous data stream using machine learning algorithms joint occurrence concept drift and imbalanced issue becomes provocative. Also, imbalance is again challenging when multi-class with minority class that too data-difficulty factors. Incremental ensemble models found promising handling theses issues. But most approaches are for two-class streams which can’t be utilized multiclass streams. this paper we have designed hybrid dynamic chunk model (HDCEM) insect-data issue. To deal proposed effective split bagging algorithm has achieved better performance on recall F-measure arriving chunks from stream. HDCEM can adapt to abrupt gradual because it combined features both online chunk-based together. It average 78% insect 71%