Exploring frequency-based approaches for efficient trajectory classification

In the last few years, several trajectory classification methods have been proposed for mobility data collected from GPS devices. Most of them only use information derived from the physical movement of the object, as speed, acceleration, and direction variation. More recently, trajectory data obtained from location-based social networks, based on user check-ins in Points of Interest (POIs), have been used to analyze user mobility patterns, giving rise to the development of methods for this specific type of data. While GPS trajectories are in general dense, and movement is characterized by spatio-temporal features and sequential patterns, social media trajectories are mostly sparse, and we claim that the moving object can be characterized simply by the frequency of visits. In this paper we propose a simple, effective, and efficient trajectory classification method based on POI frequency. With experiments on three real datasets we show that the proposed method outperforms the state of the art and is suitable for large amounts of data.