Abstract: “Mining stream data have recently garnered a great deal of attention for Machine Learning Researcher. The major challenges in stream data mining are drifting concept that deals with data whose nature changes over time. Concept drift is one of the core problems in Stream data mining and machine learning. Classification of Stream data in the presence of drifting concepts is more difficult and one of the core issue. In this paper, A Classifier based on hybrid approach is proposed and implemented that handle concept drifting stream data. The proposed classifier is used Naives Bayes as base learner for classification of concept drifting stream data where as concept drift is detected and handled by using drift detection method. Experiments and results on datasets show that the proposed approach performs well with improvement in accuracy of classification and can detect and adapt to concept drifts.”

Keywords: concept drift, stream data, classification, drift detection.