A Model of Website Usage Visualization Estimated on Clickstream Data with Apache Flume Using Improved Markov Chain Approximation

Visualization of the website clickstream data has been a pivotal process as it aids in defining the user preferences. It includes the processes of gathering, investigating and reporting about the web pages that are being viewed by the users. Clickstream visualization is primarily employed by organizations which focuses on gaining the user preferences and improve their products or services towards achieving maximum satisfaction of users. Most existing visualization tools come up short in helping the organizations achieve this goal. Markov chain model is the commonly utilized method for developing data visualization tools. However the issues such as occlusion and inability to provide clear data visualization display makes the tools volatile. This paper aims at developing a visualization tool named as WebClickviz by resolving the above mentioned issues by improving the Markov chain modelling. A heuristic method of Kolmogorov– Smirnov distance and maximum likelihood estimator is introduced for improving the clear display of visualization. These concepts are employed between the underlying distribution states to minimize the Markov distribution. The proposed model named as WebClickviz is performed in Hadoop Apache Flume which is a highly advanced tool. Through the experiments conducted on evaluation dataset, it can be shown that the proposed model outperforms the existing models with higher visualization accuracy. Key-Words: Clickstream data, Data Visualization, Hadoop, WebClickviz, Apache Flume, Markov chain, Kolmogorov– Smirnov distance, maximum likelihood estimator, heuristic approximation.