Learning Using Concept Drifts: An Overview

Abstract

The variations of concept in unceasingly evolving data streams is defined as concept drift. It is needed to address the problems raised due to concept drifts and to adapt the concept drift in existing system for enhanced performance in nonstationary environment. This can be accomplished by designing classification system such a way that, concept drift is considered and hidden, meaningful knowledge is also extracted. The paper discusses different techniques to handle concept drifts and applications of concept drift in real world.