Choice Behavior Under The Impact Of Multivariate Traveler Information

With the rapid development of information technology, the Internet and mobile Internet emerges and plays an important role in people's daily life. Travel information acquisition through the Internet and mobile internet is becoming a habit. The relationship between traveler and information is closest ever. The research on travel behavior under information impact is promising to provide useful decision support for developing advanced traveler information service system, traffic demand management policy and releasing traffic congestion.According to the literature, the impact of information on activity and travel can be summarized as long-term decision including choices of activity pattern, travel demand, mobility and residency and short term decision including choices of travel mode, departure time and route. However, current research suffers from the following two shortages. First, a great majority of the research efforts have focused on discussing behavior under provision of some specific information. Second, for those mentioned multi-variate traffic information are just taking it as the general background. Detailed description and identification of multi-variate traffic information environment is needed. Moreover, a majority of research related information and choice behavior are seeing from the trip-based perspective. A deeper angel to explain the inner mechanism of traval is required.This thesis defines the specific content of multi-variate traffic information as 7 categories including information of urban road network, traffic condition, navigation and position, public transit service, interchange, customized service and news. The framework of decision process of activity and travel choice under multi-variate traffic information is proposed. Based on that, the modeling type and research object is determined.The SP optimal orthogonal design method is developed to obtain efficient observations under the small sample situation. The essence of this method is to maximize the difference of generic attribute levels between alternatives while keeping orthogonality for both generic and alternative specific attributes within each alternative. The RP-SP fusion technique is used to generate the survey questionnaire. Finally, the travel and information usage survey is conducted in Chengdu. The answers of 200 respondents pass the data examination. The effective rate of survey reaches 72.5%. The observation number of different travel choice ranges from 1200 to 5400.By summarizing and comparing the socio-demographic characteristics, the sample is proved to be reasonably distributed. The further analysis of sample reveals that the trip distance of over 80% residences is shorter than 5 kilometers and more than 90% of trips take less than 1 hour. Smart phones are nearly full covered in population. Nearly 93% of the respondents download and install at least l travel-related application.The most popular timing of information acquisition happens before and on the way of conducting entertainment activities. Mobile is most used for obtaining information. As for specific content demand, position and route information are most favored while parking and taxi operation condition seems not wanted.To achieve a comprehensive understanding of how multi-variate traffic information effects on individual travel behavior, a joint analysis model which is capable of simultaneously accommodating information environment itself and multi traffic information contents is proposed. The key of this joint model is to create dummy variables to represent information provison, the complexity of the information environment indicator which equals to the sum of dummy variables and the inter-effect variables consisted of information attributes and the dummy variables. A trip-based travel mode choice model and a departure time choice model are constructed using this method. Both of them are proved to have a satisfied model fit.The model estimate result presents a series of findings:individual's trip mode choice is not effected by information environment. People tend to care more about some specific information like weather condition, parking time and incident report. However, contrast to that in mode choice, travelers are highly sensitive to information environment itself along with some specific information such as weather, bus waiting time, incident report and traffic management measure when making departure time choice. Cross analysis on some significant variables shows that elder people, company employees and students are most frequently changed their mind.Moreover, a trip chain pattern choice model, a trip chain-based mode choice model and a trip chain-based departure time choice model are built using the same modeling method of information impact analysis. It is noteworthy that a mixed model combining a rule-based algorithm and probability model is proposed to describe the hierarchy and linkage of the mode choices between trips under the viewpoint of trip chain.The result indicates that the multi variate traffic information environment will significantly impact how people generate a trip chain. The more traveler knows, the less chance of choosing complex trip chain pattern. Consistent with the result of trip-based model, mode choice seems to depend more on habit rather than information environment while the departure time choice is highly sensitive to it. Also, the assumption on dynamic linkage between trips is proved.