| Title | Using Cellular Data to Analyze Urban Trip Chain Patterns |
| Year | 2018 |
| Degree | Master |
| School | Department of Transportation and Logistics Management National Chiao Tung University |
| Author | |
| Summary | Trip chaining behaviors are at the heart of the transportation demand and planning, which is a phenomenon that we know exists but rarely investigate. This could be attributed to either the difficulty in defining trip chains, the difficulty in analyzing all the possible trip chain types, the difficulty in identifying their covariates, or all of the above. Household travel diary surveys are the most traditional and widely adopted approach to collect and analyze travel patterns of individuals and households. However, due to the limitation of sampling budget and time, household surveys are usually of low coverage and low updating frequency. With the rapid advance in technology, new technologies have emerged, such as Global Positioning System (GPS) and Cellular Positioning System (CPS). Thus, this study attempts to adopt GPS and CPS data to define trip chain patterns and identify their contributing factors, including trip characteristics, socio-economic attributes, and built environmental variables. Firstly, trip and trip chain based on GPS and CPS data are defined. Trip chaining behaviors are then classified into various types according to the number of trips in a trip chain. A total of four models at two levels of details (collective and individual) and on two types of data (CPS and GPS) are estimated and compared. The collective models simultaneously regress the numbers of various types of trip chains in a district on potential contributing factors by using multivariate regression models while the individual models regress the type of trip chain of a sampled user on potential contributing factors by using multinomial logit models. A case study on Taipei Metropolitan is conducted. A total of four types of trip chains are classified. The estimation results show that the estimated models reveal similar effects of selected contributing factors either based on CPS or GPS data. However, the explanatory power of GPS-based collective models is lower than that of CPS-based models, which may be due to the limited number of GPS samples in comparing to CPS samples. According to the estimated individual models, travelers with shorter average trip distance, in the higher public transportation usage area,II originating from residential area, and heading to industrial or commercial areas tend to have more trips in their trip chain. Additionally, the estimated nested logit model shows the inclusive value fails to be significantly tested, implying that the correlation among trip chain types is not significant. |
| Hashtags |
View count:
160