Analysis of World Wide Web Traffic by Nonparametric Estimation Techniques

The study of measurements of world wide web traffic has shown that different characteristics are governed by long-tail distributed random variables. We discuss the nonparametric estimation of their corresponding probability density functions. Two nonparametric estimates, a Parzen-Rosenblatt kernel estimate and a histogram with variable bin width called polygram, are considered. The proposed estimates are applied to analyze data of real web sessions. The latter are characterized by the sizes and durations of sub-sessions as well as the sizes of the responses and inter-response time intervals. By these means the effectiveness of the nonparametric procedures in comparison to parametric models of web-traffic characteristics is demonstrated.