In this paper, we examine if the Relative Neighborhood Graph (RNG) can reveal related dynamics of page-level social media metrics. A statistical analysis is also provided to illustrate the application of the method in two other datasets (the Indo-European Language dataset and the Shakespearean Era Text dataset). Using social media metrics on the world's 'top check-in locations' Facebook pages dataset, the statistical analysis reveals coherent dynamical patterns. In the largest cluster, the categories 'Gym', 'Fitness Center', and 'Sports and Recreation' appear closely linked together in the RNG. Taken together, our study validates our expectation that RNGs can provide a 'parameterfree' mathematical formalization of proximity. Our approach gives useful insights on user behaviour in social media page-level metrics as well as other applications.
12th International Conference on Advanced Data Mining and Applications (ADMA 2016). Proceedings of the 12th International Conference on Advanced Data Mining and Applications [presented in Lecture Notes in Computer Science, Vol. 10086] (Gold Coast, Qld 12-15 December, 2016) p. 283-297