A Novel Approach to Address Concept Drift Detection with the Accuracy Enhanced Ensemble (AEE) in Data Stream Mining

Data streams present a unique challenge in effectively managing concept drift, making data mining a complex task. To overcome these challenges and achieve high-performance classifiers, introduces a novel methodology called Accuracy Enhanced Ensemble (AEE). AEE enhances existing classifiers, including Accuracy Weighted Ensemble (AWE) and Accuracy Update Ensemble (AUE), to better handle concept drift in data streams. Unlike AWE, which trains a new classifier for each incoming data block and evaluates all classifiers using that block, AEE adopts a different approach. It enhances classification accuracy by assigning weights to classifier components based on their expected performance on test data. Experimental results demonstrate that AEE consistently outperforms AWE in terms of categorization accuracy. Moreover, AEE exhibits lower memory usage compared to other ensemble algorithms. However, prior research indicates that AUE may face challenges related to memory usage and processing speed during the mining process. Nonetheless, AEE effectively addresses concept drift by continuously updating ensemble classifiers, leading to improved accuracy. Further testing and refinement of AEE is necessary to optimize memory utilization and processing performance. Overall, the findings suggest that AEE shows great promise in enhancing data mining in concept drift data streams. By maintaining high performance and adaptability to changing data distributions, AEE has the potential to significantly improve the accuracy of data mining tasks in dynamic data stream environments.