Understanding Travel Mode Choice Behavior: Influencing Factors Analysis and Prediction with Machine Learning Method

Abstract

Building a multimode transportation system could effectively reduce traffic congestion and improve travel quality. In many cities, use of public transport and green travel modes is encouraged in order to reduce the emission of greenhouse gas. With the development of the economy and society, travelers’ behaviors become complex. Analyzing the travel mode choices of urban residents is conducive to constructing an effective multimode transportation system. In this paper, we propose a statistical analysis framework to study travelers’ behavior with a large amount of survey data. Then, a stacking machine learning method considering travelers’ behavior is introduced. The results show that electric bikes play a dominant role in Jinan city and age is an important factor impacting travel mode choice. Travelers’ income could impact travel mode choice and rich people prefer to use private cars. Private cars and electric bikes are two main travel modes for commuting, accounting for 30% and 35%, respectively. Moreover, the proposed stacking method achieved 0.83 accuracy, outperforming the traditional multinomial logit (MNL) mode and nine other machine learning methods.