Shoji Matsumoto

Commodity Distribution Model Incorporating Spatial Interactions for Urban Freight Movement

The commodity distribution model proposed in this paper is developed in such a way that the movement of commodities is explained as an outcome of their flow through several freight agents in a supply chain. As commodity flow is fundamentally determined by demand, the proposed model was developed from a discrete choice model that considered the individual behavior of a customer to decide the suppliers from which to purchase and the amount of commodity to acquire from each of them. The model not only takes into account the interplay between shipper and customer in a supply chain but also captures the spatial interactions among alternatives and among customers, because spatial effects generally affect customer preference. In this study, several model specifications were developed and compared, with and without incorporating spatial interactions. The empirical results of the model, which were applied to analyze the urban commodity distribution in the Tokyo metropolitan area, indicate that integrating both spa...

Modelling a social dilemma of mode choice based on commuters' expectations and social learning

This study attempts to apply an agent-based approach to modelling a social dilemma of travel mode choice considering psychological and sociological aspects. A traveller is modelled to have expectations, which shows the travellers beliefs about the influence of other group members on his action, as decision-making rules. Social interaction using a group-based interaction is hypothesized to be important. We apply an imitation game based on social learning mechanisms to the model. Two kinds of mechanism are used: payoff-biased and conformist transmission. The model reveals the conditions that make cooperation as a possible outcome are optimistic bandwagon expectations, group-based interactions, and strong conformist transmissions.

Estimating Truck Trip Origin-Destination with Commodity-Based and Empty Trip Models

This study estimates origin–destination (O-D) matrices of light and heavy trucks on the basis of the flow of commodities in the Tokyo metropolitan area. The truck O-D matrix is generally determined by either vehicle trip-based or commodity-based approaches, although the former cannot distinguish between loaded and empty trips and does not characterize the shipments. Three major concepts are proposed in this study. First, the truck trip O-D is estimated on the basis of the commodity approach because it can utilize the characteristics of the shipments. Second, the main contribution of the model is its ability to estimate both loaded and empty trips by modeling the truck movements as round trips and trip chains. The O-D of truck movement, particularly the movement of loaded trips in round trips or zero-order trip chains, is similar to that of the commodity flows. However, both movements have relatively different O-D when the loaded trips travel from one origin to many destinations or are part of an nth-order trip chain. Finally, the trip chain is modeled on the basis of characteristics such as average payload, adjacent zones, and the commodity O-D providing the most attractive zones traveled by trucks. The performance of the model is demonstrated by using the mean square error between the estimated and observed truck O-D matrices. The model concept is then applied to lightweight products obtained from the food industry. The proposed concept enhances the trip chain behavior and provides better results than the model without trip chain behavior.

Dynamic mode choice of commuters in an agent-based simulation model with inductive learning machines

This study applies an agent-based approach to modeling a transport system. Using the advantage of agentbased models of being validated at an individual level, a social dilemma of travel mode choice is modeled and viewed as a complex system. An inductive learning machine is combined with an evolutionary approach to simulate traveler learning. A user-equilibrium point predicted by conventional analysis is reached and stabilized. The stable situation is produced by interaction among agents and by behavioral change of each agent, without a central or external rule that organizes objectives of the system. The study shows conditions that may produce other stable situations besides the user equilibrium point. An emergent situation combined with traveler sensitivity to payoff differences is influential.

Microsimulation for Commuters' Mode and Discretionary Activities by Using Neural Networks

This research aims to develop an activity-based travel demand model and its microsimulation model for commuters’ work-tour mode and their discretionary activities & travel by using neural networks. The model system is designed as a series of hierarchical submodels. At the highest level is a given condition of primary travel pattern of employees between home and workplace, which constrains their behavior whether and how to make discretionary tours. At lower levels of the system are the choice of discretionary travel generation and subsequently the choice of their destination, travel mode and activity duration time. The study employed the person-trip survey data for the metropolitan area of Nagaoka, Niigata conducted by the national government in November 1999. The empirical estimation by neural networks revealed that an activity & tour were not independent but closely interrelated among a daily activity pattern. Next it simulated an individual discretionary travel pattern under a number of conditions assuming the introduction TDM measures such as flexible work times or staggered work hours. The microsimulation showed its practical capability to predict the impacts of TDM measures on daily travel patterns.

Discrete Logit Modelling Based on Stated Preference Data of the Analytic Hierarchy Process for Parking Choice

The analytic hierarchy process (AHP) is a powerful and qualitative method, that can be considered a means of collecting stated preference (SP) data. But, when it comes to its application to a discrete analysis of travel behaviour or attitudes, the AHP has some methodological limitations. This chapter proposes a modelling framework to develop discrete choice models for a group by using SP data of the AHP. The model is applied to parking choice preferences in the central area of Nagaoka in Japan. We find that absolute measurement is a consistent and flexible method to apply the AHP to the analysis of travel behaviour and preferences, and nonlinear relationships can be specified between the grades of AHP and physical attributes. To estimate the discrete logit model with better statistical indicators, we recommend employing the alternatives ranking process, where the choice set of alternatives includes a few alternatives of higher rank. The estimated discrete logit model for parking consists of two parts with good statistical reliability: the capacity of parking places by using revealed preference (RP) data, and other factors such as convenience, facility and economy by using SP data of the AHP. In comparison with the discrete choice model using RP data only, the model using the AHP and RP data can include plural and intangible factors.