Travel Behavior Models with Explicit Consideration of Information Search and Processing toward the Evaluation of Advanced Traveler Information

本研究旨在以“有限理性”概念为基础，放宽传统模型中“工具理性”的假设，提出定量评估先进旅客资讯系统影响的方法论。首先，我们建立了两个离散选择模型，该模型明确考虑了信息搜索和加工的机制：一个是关于信息获取的，一个是关于信息加工的。潜在类模型被用来表示个体间决策过程的异质性。对这些模型进行了实证应用，结果表明，与假设“工具理性”的传统模型相比，所提出的模型在重复性和预测精度上都有明显的优势。同样得到验证的是，由于所提出的模型的自由度非常高，因此收集和利用各种d…是必要的更多的数据来源，如心理数据和SP数据，结合实际的选择数据，得到无偏有效的参数估计。其次，研究了先进的出行者信息系统对交通网络需求的影响。建立了多等级用户均衡模型，其中各等级之间对出行时间和路网结构的认知水平不同，并将其应用于名古屋大都市区的公路网。结果表明，当我们设置20%-30%的司机，主要是私人出行司机，只选择主干道时，我们得到了与观察到的交通流最好的拟合。这在主干道上造成了不必要的拥堵。通过提供完整网络的信息来实现平衡状态，这些司机将获得约70,900日元(600美元)的出行时间减少收益。但我们不能得出结论，这种对主干道的集中是由于路网结构信息的不完善，或者仅仅是因为对主干道的偏好偏差。再次，通过在名古屋大都市区进行的大规模试车实验，对试车数据进行了分析，以直观地观察路径选择行为。我们研究了一条收费路线和几条免费路线可以使用的频繁出行的O-D对。调查发现，30%-70%的出行者使用收费路线，这意味着VOT分布。我们观察到，使用了多条免费航线，其中旅行时间相差10%-20%。还观察到，收费路线往往较慢，但许多司机选择了它。因此，相当多的司机做出了看似不理性的决策。这些观察结果可能部分是由不完全信息和旅行时间和成本以外的其他因素解释的。研究结果表明，与传统的问卷式出行调查相比，探索车数据在数量和质量上都可以更加丰富。较少

The aim of this research is to propose the methodologies to quantitatively evaluate the impacts of advanced traveler information systems based on the concept of "bounded rationality" which relax some of the assumptions of "instrumental rationality" postulated in conventional models. Firstly, we developed the two discrete choice models which explicitly consider the mechanism of information searches and processing: one is about the information acquisition, and the other is about the information processing. Latent class models were utilized to represent the heterogeneity of the decision making processes among individuals. These models were empirically applied, and the results showed that the proposed model was superior in both reproducibility and predictive accuracy compared to the conventional models in which "instrumental rationality" is assumed. Also verified was that since the degree of freedom of proposed models are extremely high, it is essential to collect and utilize the various d … More ata sources, such as psychological data and SP data, together with the actual choice data to get the unbiased and effective parameter estimates.Secondly, we investigated the impacts of advanced traveler information system on transportation network demand. The multi-class user equilibrium model, in which the cognitive levels on travel time and network configuration differ between each class, was developed and empirically applied to the road network of Nagoya metropolitan area. The result showed that we obtained the best fit to the observed traffic flows when we set 20-30% of drivers, mainly private trip drivers, choose only arterial roads. This causes unnecessary congestion on arterial roads. By providing information of the full network to make the equilibrium state, about 70,900 JPY (600 USD) of travel time reduction benefits accrues to such drivers. But we cannot conclude that this concentration to arterial roads is attributed to imperfect information on network configuration or simply due to the preference bias to arterials.Thirdly, probe car data are analyzed both to directly observe route choice behavior using a large scale probe-car experiment conducted in Nagoya metropolitan area. We investigated a heavily traveled O-D pair for which a toll route and several free routes can be used. It was found that 30-70% travelers used the toll route, implying that VOT distributes. We observed that multiple free routes were used among which travel times differed 10-20%. Also observed was that the toll route was often slower but many drivers chose it. Thus, substantial number of drivers made seemingly irrational decision making. Those observations may partly be explained by imperfect information and other factors than travel time and cost. We also investigated the characteristics of probe car data and its potentials toward the transportation planning, and the results showed that probe-car data could be much richer both in quantity and quality compared to traditional questionnaire type travel surveys. Less