Visual analysis of bi-directional movement behavior

The availability of massive volumes of trajectory data has made it convenient for the study of different types of movement behaviors. Among them, bi-directional movement behaviors exist ubiquitously in our daily life, from urban traffic to animal migration, and from sports to wars. To analyze bi-directional movement behaviors, people need to compare movements in two directions simultaneously for detecting similarities or differences in the movement patterns. If the movement involves tens of thousands items like vehicles or bird migration during a ten-year time span, the comparisons need to be done at both macro level and micro level. Due to the complexities of data and the challenges of analytical tasks, visual analytics is often used to take full advantage of machines' computational power as well as human's domain knowledge and cognitive abilities. In this paper, we present a comprehensive visual analytics system with three major visualization modules, including Global View, OD-pair Flow View and Isotime Storyline View, to depict bi-directional movement behaviors in a novel way, which enables a three-level exploration to help users gain insights into both macro and micro patterns. Quantitative analyses (e.g. movement model construction, modular Dol specification and key node extraction) and intuitive visualizations (e.g. parallelized flow map, bidirectional storyline chart with contour map and multi-layer heat map) are integrated into our system to provide an efficient and intuitive solution to the analysis of bi-directional movement behaviors based on big movement data. Case studies with two real-world datasets and expert interviews are carried out to demonstrate the effectiveness and usefulness of our system.