animating flows

time-lapse visualization of mobility patterns

Porto buses are equipped with a Mobile Wi-Fi service from Veniam, which provides free Wi-Fi to bus passengers. The system stores the start and end locations and times of Wi-Fi sessions, allowing to infer mobility patterns from Wi-Fi usage. The video below shows these mobility patterns. Locations were clustered with K-means, highly decreasing the complexity of the mobility graph and allowing to understand the major flows in the city.

Clustering locations enables to understand the major patterns, but detail is lost in the way. Another approach for visualizing mobility flows is illustrated in the video below. Here, instead of clustering locations, the origin-destination pairs are bundled. Edges connecting the start and end location of a Wi-Fi session are deformed and grouped in order to identify major flows. These edges do not represent the paths (they are not mapped to roads) but flows in the city. In this version, flows that have different direction avoid each other while flows that have similar direction are attracted together. A blue to red gradient is used to illustrate flow, with flows starting blue at the origin and red at the destination.

Different criteria can be explored to bundle edges together. Some examples are illustrated in the video below here flows and their coloring are based on the trip’s distance or on the destination of the trip, for example.

Finally, and mainly for fun, we can do the animation in 3D and play with the height of the flows. In the next video, the direction of the flows is ignored, and height is calculated from the density of flows, creating interesting visual effects.