Authors: Makoto Abe
Addresses: Graduate School of Economics, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Abstract: Utilising data from a log file, a model for step-ahead web page prediction that permits adaptive page customisation in real time is proposed. It predicts the next page of a viewer based on a variant of a Markov transition matrix that is computed from page views of other visitors who read the same pages as that viewer did thus far. The prediction model accounts for viewer heterogeneity by obtaining information from visitors who visited the same pages, an idea similar to collaborative filtering in recommendation engines. The empirical result with a test site implies that it is sufficiently accurate (50.3%) in predicting page transitions. Prediction of a site exit was even better with 100% of the exit predictions and 90.8% of the continuation predictions being correct. The result was compared against other models, such as logistic regression, decision tree and neural networks, for predictive accuracy.
Keywords: web page transition; prediction models; log files; statistical modelling; Markov transition matrix; data mining; adaptive page customisation; collaborative filtering; predictive accuracy; electronic business; e-business; personalised marketing; e-commerce; electronic commerce.
International Journal of Electronic Business, 2005 Vol.3 No.3/4, pp.378 - 391
Published online: 30 Jun 2005 *Full-text access for editors Access for subscribers Purchase this article Comment on this article