Identifying Travel Mode Using GPS Based Travel Survey Data

With the further development of socialist modernization, people’ living standard is increasing high. And the number of vehicles has increased. As a result, urban transport is facing increasing pressure, which causes traffic congestion, environmental pollution and other issues. Conducting resident travel survey and analyzing travel behavior as well as mastering residential travel regularity and travel characteristics are important measures to ease traffic congestion and other problems.Among the resident travel survey, traditional OD survey is the most widely used, and its core idea is to get travel information through household survey and other ways. However, this kind of survey is subject to respondents. In real investigation, respondents often forget or slip some trips, which leads to poor accuracy to survey data. Thus, there are some demerit in traditional OD survey, such as low efficiency, poor accuracy and high burden. As a new survey tool, GPS device is featured with fast and accurately recording personal travel data, attracting wide attention of scholars. For GPS travel survey, respondents’ subjective perception cannot influence survey data. And GPS logger can record data accurately, which improves the reliability and integrity of survey data significantly. Also, GPS travel survey has some advantages, such as high survey efficiency, high data accuracy, getting a large amount of information, and etc. GPS travel survey has become an effective way to get information about residential travel behavior. Processing data and fetching travel information from the trajectory data to identify travel mode is the new way to analyze residential travel behavior.This paper presents a hybrid algorithm for mode identification by using large-scale GPS survey data. This hybrid algorithm includes subway detection and hierarchical mode identification model(Nested Logit model). The hybrid algorithm can detect the modes of walk, bicycle, subway, car and bus. Firstly, since subway travel routes have significant certainty, this paper takes spatial distance parameter threshold into account to identify subway. Meanwhile, in order to compare the accuracy of subway detection method, subway identification model based on multi-nomial Logit(MNL) is also established to identify subway and non-subway. Then, because there are differences between the various travel modes, the modes of walk, bicycle, car and bus were determined by establishing a Nested Logit model. Comparison of the identification results reveals a relatively higher accuracy provided by the method of subway detection by considering spatial distance parameter threshold. The Nested Logit model is also of high accuracy. The results indicate that the algorithm shows a high level of identification accuracy.The travel mode identifying algorithm in this paper is an effective method to detect modes of travel. The results can be used to identify travel modes and analyze residential travel characteristics based on GPS survey data, which will significantly enhance the efficiency and accuracy of travel survey and data processing. The conclusion is of significance to promote GPS in resident travel research area, analyze residential travel regularity, diagnose traffic problem, ease traffic congestion and to develop transportation system.