Using page-level metrics of a randomly selected group of 15,625 among the top 100,000 Facebook check-in locations which rank high in terms of customer engagement, we explore if the short-term dynamical information on these metrics could deliver, via a clustering approach, some new insights for marketing decision making. Using a highly-scalable clustering algorithm, statistical methods, and combinatorial optimization metaheuristics based on memetic algorithms, we have observed that some pages naturally cluster with others that share the same user-defined category. Our results highlight the need of suggesting other further " meta-categories " that encompass several user-reported categories for pages and that a priori geographical segmentation might be necessary to investigate more relevant patterns that take into account seasonal variability of behaviours and physical proximity.
Lucas, B., Arefin, A. S., Vries, N. J. D., Berretta, R., Carlson, J., & Moscato, P. (2015). Engagement in Motion: Exploring Short Term Dynamics in Page-Level Social Media Metrics. In 2014 IEEE Fourth International Conference on Big Data and Cloud Computingdoi:10.1109/bdcloud.2014.56