Improvised Feature Selection Method via Global Minimization Techniques

Abstract

The amount of information over internet has been growing last few years. And it has caused risk of information problem of accessing related data to the users. The information demand of the online users can be figured out by evaluating user’s web navigation behavior. Web Usage Mining (WUM) is used to extract knowledge from Web users access logs by using Data Mining Techniques. One of the applications of WUM is Web Sites Recommendation system which is personalized information filtering technique used to either determine whether a certain user will approve a given item or to identify a list of items which can be of significant importance to the user. In this paper the modified architecture that integrates item information with user’s access log data and then find pattern and make pattern clustering. There after generates a set of recommendations for the user. So execution time and fetching time is reduced. Other experiments compared the CFS to a wrapper - a well-known approach to feature selection that uses the target learning algorithm to evaluate sets of features. In many cases CFS has given results comparable to the envelope, and in general, surpassed the envelope on small sets of data. CFS runs much faster than the wrapper, enabling it to extend to larger sets of data.