Origin-Destination Trips for Human Mobility Based on Twitter

The availability of massive anonymized datasets provides a novel insight for understanding human mobility which is important to disease prevention, city planning and traffic forecasting. Most relevant studies employ call detail record (CDR) data with convenient collection but low location resolution. Twitter data, as an alternative, can be located within ten meters. This paper focuses on studying the human mobility with Twitter data in Milano, Roma and Venezia. Firstly, the stay points of Twitter users are extracted and classified into three types: Home, Work and Other. Then, Grid-based cluster algorithm is proposed to clustering these stay points into different stay regions. Next, the trips of Twitter users are divided into three categories: Home-Based Work (HBW), Home-Based Other (HBO) and Non-Home Based (NHB). Finally, the Twitter users’ trips are plotted and visualized by Geographic Information System (GIS). Computational results explicit that the tendency of trips length and times are in accordance with Zipf’s law.