Visual Analysis of Predictive Suffix Trees for Discovering Movement Patterns and Behaviors

The use of GPS-equipped devices has allowed generating and storing data related to massive amounts of moving objects, promoting many solutions to movement prediction problems. Movement prediction became essential to perform tasks in several areas ranging from analysis of the popularity of geographic regions; and management of traffic and transportation; to recommendations in location-based social networks. To explore this type of data is a complex task because one must deal simultaneously with space, time and probability. In this work, we apply the branching time concept to visual analytics, proposing an approach that supports movement prediction using Probabilistic Suffix Trees. We try to substitute the traditional evaluation method, based on reading texts, by an interactive visual solution. To validate the proposed solution, we developed and tested a visualization tool using a real dataset. It assisted experts to quickly identify where a person lives, where she works and to recognize some of her movement patterns and probable behaviors.