Based On The IC Card And GPS Data To Infer The Type Of Travel Activity

With increasing the popularity of IC card fee collection system (ICFC) and automatic vehicle location system (AVL) in transit system, many meaningful information like passenger travel pattern, origins and destinations can be derived. However, on the one hand, the absence of trip purpose and socio-economic characteristic is an intrinsic limitation of automated fare collection data. On the other hand, the high-density development and mixed land use of urban in China cause the diversity trip purpose of transit passengers. Deriving trip purpose may be significant and gives us insight into travel behavior. This paper proposes a way to infer passengers’trip purpose using a typical weekdays ICFC and AVL data in Chengdu. Firstly, this paper choose adult loyalty and student loyalty in transit system to analyze their temporal and spatial travel patterns. Secondly, K-mean cluster method is used to develop heuristic rules for applying trip purpose assignment process. At last, a decision tree-based classification technique is conducted to determine the performance of the trip purpose inference model with 3 different rules. The results shows that the trip purpose assignment process with rule 3 is the most effective model to infer passenger trip purpose. The most useful independent variables for work-related activities of adult loyalty and school-related activities of student loyalty is the first transaction time. And land use of alight is more significant than travel distance and land use diversity of alight stop for inferring school-related activities of student loyalty.