Clustering Based Outlier Detection Using K-Means Strategy

The process of detecting outliers is a surveillance that comes into view to move away patently from other surveillances in the model. This arrangement is planned to demonstrate the fixations happened among the explorations occurred between client, and server. In the practical scenario all the individuals obviously are familiar with the procedure of how to transfer a request for the meticulous requirements, and how to get a comeback for that demand. On the other hand no one knows about the inside process of searching information from a huge database. Clustering is one of the best known techniques to maintain the information efficiently into the database. Clustering employs grouping of similar objects (similarity in terms of data content or there may be any other factors also). Outlier detection is one of the main divisions of data mining and deserves further research attention from data mining community. The brilliant technique for text classification process is called Feature Selection. These processes merge with k-means and produce more effective result. Words in the feature vector are grouped and forming a header to that group based on the similarity test. Each cluster is formed based on the behavior of the text with other text and the average mean value. Same words into the cluster are grouped together and produce better data maintenance as well as through this process the data searching by the user is also categorized and fledged in a probabilistic analytical manner. This paper primarily focuses on comparing various outlier detection methods based on clustering and association rule applications and also prove that this present approach is efficient enough to find the outliers and represent the outlier as the cluster head.