Model-Based Biclustering of Clickstream Data

• Activity Number: 614
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Volodymyr Melnykov*+
• Companies: University of Alabama
• Keywords: clickstream ; finite mixture model ; model-based clustering ; biclustering ; model selection

Abstract

Navigation patterns expressed by sequences of visited web-sites can characterize the behavior and habits of web-users. Such web-page routes taken by individuals are commonly called clickstreams. Clustering clickstream sequences is challenging since one needs to group categorical data sequences rather than vectors and the majority of traditional clustering algorithms are not applicable in this setting. We consider model-based clustering relying on the mixture of first order Markov models. Since the number of distinct web-pages, and therefore the number of states in a Markov process, can be high, such a mixture model involves a large number of parameters. We propose grouping states by their similarity to reduce the number of parameters in the model. The developed methodology is illustrated on synthetic and real datasets with good results.