Dynamic Real-time Classification of Data Streams

Abstract

To develop real time classification from high throughput of data stream (dynamic data) is one of the most challenging areas of big data analysis. In this proposed system we are using concept drift. (Changes of the pattern encoded in the stream over time). And imposes unique challenges in comparison with real time classification data mining from dynamic data. Several real-time classifications of data stream algorithms exist. The proposed system highlights the Fixed Width clustering, variable width calculation and global width clustering for data stream classifier. The result of these algorithms provides high accuracy, less time & high speed.