Empirical analysis and improvement of density based clustering algorithm in data streams
Madhu Shukla, Yogeshwar Prasad Kosta · 2016
Data mining has gained much importance in the field of research these days. It makes perfect blend for analyzing data of any fields and provide decision based output. Data generation and storage these days are done at high speed. Non stationary systems play holistic role in providing such data. Availability of such data creates scope of analysis for researchers. Such data which are continuous, unbounded, fast are termed as stream data. Clustering is the best method for analysis of stream data. As labeling of data is not possible for streams so clustering may assist this process. Also framing clusters digs out the points which do not seem to be part of cluster thus assisting outlier detection also. Different kinds of clustering algorithm exists but density based method helps in detecting clusters of arbitrary shape which other algorithm does not do. We have discussed an approach and presented its result in comparison of existing algorithm in this paper.