Predicting Web Site Access: An Application of Time Series

Giuliano Antoniol, Gerardo Casazza, Giuseppe A. Di Lucca, Massimiliano Di Penta, Ettore Merlo

Abstract: The Internet and Web pervasiveness are changing the landscape of several different areas ranging from information gathering/managing and commerce to software development. This paper presents a case study where time series were adopted to forecast future Web site access. In order to measure the applicability of time series to the prediction of Web site accesses, an experimental activity was performed. The log-access file of an academic Web site (http://www.ing.unisannio.it) was analyzed and its data used as test set. The analyzed Web site contains general information about the Faculty of Engineering of University of Sannio at Benevento (Italy). Preliminary results were encouraging: the average number of connections per week could be predicted with an acceptable error.