Visualizing Clickstream Data with Multidimensional Scaling
J. Azaria, Michiel van Wezel, Wim Pijls · RePEc: Research Papers in Economics · 2005
Abstract. We visualize a web server log by means of multidimensional scaling. To that end, a so-called dissimilarity metric is introduced in the sets of sessions and pages respectively. We interpret the resulting visualizations and find some interesting patterns. 1 Introduction. This paper describes the investigation of the data of a Web server log, as part of the ECML/PKDD 2005 Discovery Challenge [1]. The Web server data comprise a listing of page requests. In each request a session-id and a page-id can be iden-tified. A set of requests corresponding to the same session-id defines a session.