Data stream classification is a challenging task. For real-time, data concepts of instances keep varying with time, such as weather prediction or intrusion detection etc. Classification is a supervised technique to mine useful patterns from these real data. There is a need to mine knowledge from these large data streams with techniques those provide accurate results and memory efficient. Using a set of classifiers i.e. ensemble classifier than a single one, proves to be more efficient. In this paper, Revised Accuracy Updated Ensemble(RAUE) classifier algorithm is proposed. In this algorithm, the data streams are processed in blocks. The ensemble used in RAUE uses Hoeffding Option Trees(HOT) as basic classifier set. The experimental evaluation and comparison of the proposed RAUE with existing ensemble algorithms is carried out. Out of all the compared algorithms, RAUE shows increase in accuracy and also uses less memory for identifying class labels for instances.