An MCDM approach towards handling outliers in web data: a case study using OWA operators

World Wide Web has emerged as one of the primary modes of information sharing and searching. Its reach has been extended to daily aspects of our life whether it is related to business or education. As the information is going online, and so is the complexity of finding the correct, precise and appropriate information. Many online companies rely heavily on analysis of web data to stay in business, to make strategic decisions, and for their existence. One of the problem in analyzing the web data is the web user. A typical web user exhibits highly uncertain pattern of web browsing and the same is captured in form of web server logs. Various data mining techniques like regression, are used to analyze such kind of data, but the inherent complex nature of web data introduces some outlier values while mining for information. Minimizing these outliers has always been a challenging task for data scientist and researchers. This paper uses an aggregation-based approach based on various ordered weighted averaging operators to reduce the outlier values in regression analysis. In this paper, a regression problem is being formulated followed by solving the problem with the help of concepts of multi-criteria decision making. Results, thus obtained are able to show that outliers can be reduced to a significant amount with the help of this approach.