A non-stable definition of classes in incoming data is known in literature as the concept drift. In this paper, we focus on the topic of adaptive ensembles that generate component classifiers sequentially from fixed-size blocks of training examples called data chunks. In such ensembles, when a new block arrives, existing component classifiers are evaluated and their combination weights are updated. Compared to AUE1, we put forward a new weighting and updating mechanism as well as modify many other construction details to reduce computational costs and improve classification accuracy. In this we evaluate the proposed AUE2 algorithm with respect to classification accuracy, processing time, and memory costs. Generally, data streams can be processed either incrementally by single examples st (as described above) or they are divided into equally sized blocks (data chunks) B1, B2,…, Bn and the evaluation or updating of classifiers is performed after processing all examples from a block. One of the important challenge in the data stream is concept drift. In this paper, we propose a new data stream classifier, called the Accuracy Updated Ensemble (AUE2), which aims at reacting equally well to different types of drift. AUE2 combines accuracy-based weighting mechanisms known from block-based ensembles with the incremental data. AUE2 provided best average classification accuracy while proving to be less memory consuming than other ensemble approaches.