Weighted order-dependent clustering and visualization of web navigation patterns

This study introduces a simple but effective visualization system that allows decision makers to easily identify groups of visitors with different sequential navigation patterns. In particular, navigation sequences of visitors are encoded as an order-dependent format so that early visited pages have more weights in the clustering process. Experimental results on a real-world dataset show that Markov state-transition diagrams with transition probabilities based on the proposed scheme can be very useful for developing Web marketing programs tailored to visitors' preferences and interests.