Handling Gradual Concept Drift in Stream Data

Data streams are sequence of data examples that continuously arrive at time-varying and possibly unbound streams. These data streams are potentially huge in size and thus it is impossible to process many data mining techniques (e.g., sensor readings, call records, web page visits). Tachiniques for classification fail to successfully process data streams because of two factors: their overwhelming volume and their distinctive feature known as concept drift. Concept drift is defined as changes in the learned structure that occur over time. The occurrence of concept drift leads to a drastic drop in classification accuracy. The recognition of concept drift in data streams has led to sliding-window approaches, instance selection methods, drift detection, ensemble classifiers. This paper describes the various types of concept drifts that affect the data examples and discusses various approaches inorder to handle concept drift scenarios. The aim of this paper is to review and compare single classifier and ensemble approaches to data stream mining respectively and propose a methodology towards its contribution successfully.