Detecting Jam Regions Correlations and Predicting Taxi Transportation Flow and Velocity

Abstract

Nowadays, taxi is one of the most popular transportation modes. There is a large amount of commuter using taxi every day and taxi trajectories represent the mobility of people. In the big cities, taxi is equipped GPS device and run during 24 hours per day, they may be used to extract reliable information for transportation status. This paper states our method using taxi trajectories in Hanoi, Vietnam during 4 weeks from September 18th to October 15^th. In our method, Hanoi map is divided into the smaller regions with a predefined size. Next, we identify the contiguous regions where jams happen during different time slots and their correlations. Finally, we develop a model predicting taxi transportation flow in each region and velocity basing on historical and weather data.