A Study on Thai Internet’s Clickstream Data Using Topological Data Analysis

Social network plays an important role in our everyday life. Researchers mostly study and explore the dataset to get knowledge about its nature to create proper hypotheses. In most case, they use statistics to describe relations of data attributes. The number of experiments usually depends on the number of possible attribute relationships. However, the number of attributes, as well as the number of observations, in social network data are generally large so that statistical method might not be a suitable tool for the study. In this paper, we propose to explore a visual interpretation tool, called Topological Data Analysis (TDA), to analyze relationships of attributes in a Thai Internet forum (Pantip.com). We mainly focus on users’ reading and sharing behaviors extracted from the clickstream data. The results show that TDA can help spot the relations of attributes, and we can use these insights to better understand the nature of Pantip users’ communities.