Figure 2

Average uniqueness of movement information. For each dataset, we measure the average uniqueness of the users’ movements (y-axis) as we vary the number of movement points considered (x-axis). Specifically, each movement point registers the average distance (red), speed (blue) or direction (green) of travel over a time window of 30 seconds. As expected, the average uniqueness increases as we add more movements points. The effect varies over the three datasets, but in all of them the average direction of travel appears to be the most discriminative feature.