Transportation modes behaviour analysis based on raw GPS dataset

Abstract

Significant information exists in the global positioning system (GPS) data for understanding behaviours and transport planning. However, fine-grained identification of transportation modes is still required. In this paper, we present a robust framework to identify different means of transportation modes from raw GPS dataset. We make the following contributions: 1) we design an effective trajectory segmentation algorithm to divide raw GPS trajectory into single mode segments based on logical assumptions; 2) we propose several modern features, which are more discriminating than traditional features; 3) we adopt an additional segments expansion procedure by considering the wholeness of trajectory. Experiments prove that our framework achieves a promising accuracy for identifying transportation modes.