Search

For today's data savvy information systems, it becomes inevitable to adapt stream-processing techniques where sequence of data items arrives continuously over the period. Streaming data analytics is thus an emerging area of research where standard data mining is insufficient to produce desired efficacy. This paper aims to focus on streaming data classification and presents a comprehensive spectrum of various ensemble approaches, through systematic experimental work taking the stream data both from artificial and real time data. The paper also contributes to provide performance evaluation of existing novel and adapted algorithms applied to artificial and real time streams, with the intension to provide the research community with the classifier ensemble techniques tuned for artificial and Real time data separately. Authors of this paper would like to address real time streaming analytics majorly, as it needs more attention from research community since there is still scarcity of established benchmarks and standardized frameworks. The paper concludes with summarization of major observations in evaluation experimentation and discussion of open research directions for future work. [ABSTRACT FROM AUTHOR]