Linear Approximation of Web user Categorization

Web logs are being utilized as huge data repository for mining interesting and potentially useful patterns. When mined properly these patterns provide the support to designer of the web site; analyst and management executives in strategic decision making such as improvement of content, structure or in making adaptive web sites. Web logs although contains many information but doesn't clearly indicate the page refreshing. If we are able to record the refreshing then in the Markov chain model we can add one more state which can be utilized to categorize the users of website into three categories i.e. faithful, Partially Impatient and Completely Impatient users. Jain et al. [3] in his paper derived some theorem to study each type of users’ behavior and shown that how do users behavior differ. In this paper we approximate the expression using straight line and least square method and did comparative study. We find that approximate expressions can be used to predict the behavior of the users which is independent of parameters also.