Detecting Outliers in Data streams using Clustering Algorithms
The data stream is a new arrival of research area in data mining where as data stream refers to the process of extracting knowledge structures from nonstop, fast growing data records. Emerging applications involved in data streams are motivated by many researches involving continuous massive data sets such as customer click streams, ecommerce, wireless sensor network, network monitor, telecommunication system, stock market and meteorological data. For handling this type of large data, the current data mining systems are not sufficient and equipped to deal with them, for this cause it leads to a numerous computational and mining challenges due to shortage of hardware limitations. Nowadays many researchers have focused on mining data streams and they proposed many techniques for data stream classification, data stream clustering and finding frequent items from data streams. Data stream Clustering and outlier detection provides a number of unique challenges in evolving data stream environment. Data stream clustering algorithms are highly used for detecting the outliers efficiently. The main objective of this research work is to perform the clustering process and detecting the outliers in data streams. In this research work, two clustering algorithms namely CURE with K-Means and CURE with CLARANS are used for finding the outliers in data streams. Different sizes and types of data sets and two performance factors such as clustering accuracy and outlier detection accuracy are used for analysis. By analyzing the experimental results, it is observed that the proposed CURE with CLARANS clustering algorithm performance is more accurate than the existing algorithm CURE with K-Means.
Dr. S. Vijayarani Ms. P. Jothi