This paper presents two aggregate models of destination choice which incorporate probabilistic formation of a choice set, in order to make possible us to forecast trip distribution efficiently. We assume that an alternative is dependent of the others and is selected into a choice set only if the utility exceeds a stochastic threshold. Therefore, we formulate alternative-selection probability model. The one is logit model in which the utility function is truncated by alternative-selection threshold. The other model is based upon the idea of importance sampling alternatives, but the sampling probability is replaced by alternative-selection probability. The result obtained by empirical study shows that both our models are statistically significant and can reproduce trip distribution observed very well.

View full abstract

The purpose of this study is to discuss the effect of rainfall, the shape of a convexity and the density of convexities on the visibility of raised-rib road marking on a rainy night, and the characteristics and the effect of raised-rib road marking were confirmed.
The summary of the obtained new results from this study is below. Firstly the convexities do not always improve the retro-reflection from raised-rib road marking on a rainy night. Secondly in order to get retro-reflective coefficient from the slope of a convexity more than marginal value it is important to keep the slope degree of a convexity. Thirdly it is effective against improving visibility to increase the weight of sprayed glass beads on the slope. Finally the height of a convexity should be determined according to rainfall, the density of convexities and so on.

View full abstract

This paper describes a model to predict cyclists' behavior of choosing the route in a urban road network. The model is based on the assumption that individual cyclists choose their route to minimize the sum of disutilities derived from the road and traffic conditions of network links. Specifically speaking, a disutility of a link is a function of its length, number of intersections, gradient and an average disutility of a passage. Since a passage is composed of a roadway and a sidewalk, the average disutility is derived from each disutility and the rate of cyclists choosing a sidewalk, A field study has indicated that the presented model produces small difference between the calculated values and measured one.

View full abstract

This paper compares the dynamic flow patterns on two symmetrical networks: “Evening-network” which is a network with a one-to-many OD pattern; “Morning-network” which is one with a many-to-one OD pattern that can be obtained by reversing the direction of links, an origin and destinations of the Evening-network. Although the conventional static traffic assignment produces exactly the same flow pattern for both of the networks, we nevertheless find that such a simple conclusion does not hold for dynamic assignment models. As an example demonstrating the essential difference between the two dynamic flow patterns, a capacity increasing paradox (dynamic version of Braess's paradox) is also discussed

View full abstract

Among more than two disaggregate behavioral data like revealed and stated preference data collected from same individuals, the serial correlation is usually included. This paper develops a discrete choice model and its estimation method that combined some serially correlated datasets. This method is applied to SP and RP data for the introduction of P & BR system and time series location data of same lots. These two case studies show effectiveness and availability of the proposed method.

View full abstract

The objective of this research is to find factors that affect our recreational behavior by using disaggregate models for such trips. More specifically, a discrete-continuous model is proposed to explain the choice of the number of days spent for recreational trips in a year and the type of the trips. The model is applied to the national survey of recreational behavior to find that regional economic and recreational resources substantially affect the behavior.

View full abstract

A game theoretic approach is applied to the net benefit allocation problem in reservoir renewal projects. Since renewal projects are implemented at the sites where existing agencies have already used reservoirs, the problem of allocating net benefit as required by renewal projects is more difficult than that of simply allocating costs in the case of constructing a new reservoir. In this paper, the differences in bargaining power between new and existing agencies are reflected explicitly in the permission stricture between users and owners of sites. Then, the net benefit allocation method based on the permission structure is proposed. The modified allocation method including the time limit on user's right is shown to be more reasonable when we account for allocating water resources for a longer span of time.

View full abstract

This paper is aimed to develop a new road surface management system (RSMS) that is simple, practical and affordable for developing countries to adopt it. The method of evaluating pavement surface based on visual inspection is first proposed. The pavement condition index (PCI) is adopted and modified to be flexible so that any particular circumstances in developing countries can be added. A pavement performance model is, then, established to predict road surface distress condition, and used as a basis of estimating pavement maintenance costs. A case study of applying RSMS to Bhutan is conducted to show that RSMS can properly serve decision-makers as an effective tool for road surface management.

View full abstract

In recent years, researchers have paid a lot of attention to Layered Neural Networks (LNNs) as a non-parametric approach for the classification of remotely sensed images. This paper focuses on the generalization capability of LNNs, that is, how well an LNN performs with unknown data. First, we clarify its description from the point of view of information statistics. With this discussion, we provide a feasible technique to design the LNN in consideration of its generalization capability. Finally, we apply the proposed technique to a practical land cover classification using remotely sensed images, and demonstrate its potential.

View full abstract