Recently, new approaches have been tried and applied in the conduct of travel behavior surveies. Notable of these approaches are in the collection of data and analysis particularly in the field of travel behavior and mobility. Two of the most prominent survey approachs are Activity Based Approach and Panel Survey. These approachs will furnish pertinent information in supporting the analysis of a range of problems which ordinarily could not be addressed by conventional cross-sectional data.
Traditional travel demand models are estimated from revealed preference data with manifest explanatory variables and assume that the choice set is a priori given.This paper presents a methodology that combines revealed preference data and psychometric data in the development of travel behavior models. The approach is based on a theoretical framework that includes latent psychological factors of attitudes, perceptions, preferences and choice sets. General formulation of the framework is presented followed by more detailed presentation of three submodels: combined estimation from revealed and stated preference data, discrete choice models with latent explanatory variables, and development of latent choice set models.
On the context of transportation planning, one travel behavior must be usually understood as a process including some decision makings. This paper describes discrete choice analysis of individual shopping travel behavior which consists of three dimensional travel decisions, i. e. shopping frequency, shopping destination and parking place choice. For empirical analysis, discrete choice models are applied, which are based upon random utility theory in microeconomics. We assume 3 level-nested structure for the decision makings and apply Nested Logit model and/or Ordered Logistic model, since the I. I. A. propetrty of logit model is not suitable for3dimensional travel decisions.The result of empirical analysis shows that the assumption of choice structure is statistically significant and O. L. model of frequency choice is very efficient. So, the relevant framework of travel behavior model is successfully developed. Finally, in order to represent validity and usefulenss of the model, some transportation polices are simulated and evaluated.
Dynamic route guidance systems are expected to achieve smooth and safety traffic and improve road traffic environment. It needs to prepare a prediction model of traffic condition, in which driver's route choice behavior is explicitly involved. Dynamic traffic assignment models have been studied for describing road network flow, however, road transport informatics(RTI)such as route guidance systems are not considered in those models. This paper shows the relation between network assignment models and RTI in order to identify the roles of traffic assignment models for constructing dynamic route guidance systems and to discuss necessity conditions for model development.
The number of traffic accidents has been increasing in the recent years in Japan, while the number on other advanced countries trends decreasing. The effective policies for traffic safely had been conducted in1970's and there are no ideas which can decrease the number of traffic accidents in a short term. Therefore, from a viewpoint in along term, new infrastructure planning for traffic safety is required. In this paper, the relations between road structure and traffic accident are studied. A Geographic In formation System(GIS)is built for the analysis of traffic accident which is caused by many factors. First, it is shown that the visualization by the GIS is effective for accident analysis. Second, several multi-variate analyses with the data processed from the GIS are examined and the factors of accidents are specified. Finally, the required additional data and analysis methods for further research are discussed.
This paper presents a basic concept of transportation planning in recreational areas where the arterial roads tend to have serious traffic-congestion in the peak-hours on non-workdays. Based on some characteristics of traffic flow patterns in those recreational roads, a simulation model of travel time is developed for the purpose of quantitatively describing the effect of cardrivers' parking behaviors on the reduction in the speed.
Dichotomous data can be analyzed in two different formats; the item-category form and the multiple-choice form. These two forms, as analyzed by dual scaling, typically lead to different results.This discrepancy offers an interesting situation, from which several hidden properties of dual scaling may be identified and derived, in particular, the relation between dual scaling and principal component analysis. Through the exploration of this relation, a number of mathematical properties of dual scaling weights and scores can be derived. Some numerical illustrations of interesting points are presented.
In risk theory, utility functions are usually assumed to be differentiable everywhere. In this paper, this assumption is weakened, and utility functions are allowed to have kinks. A new measure of risk aversion is proposed, and is used to generalize Theorem1in the weak form given in Pratt(1964).
This paper proposes a goods evaluation model by introducing perceptual encoding. We first distinguish between physical characteristics and attributes by dividing information on goods, acquired by consumers, into information on the physical characteristics and that on the situations in which goods are used. That enables us to define perceptual encoding as that consumers recognize the physical characteristics and evoke the situations to estimate the attributes by the law of cause and effect they have learned through purchase experience.Based on the definition, we build a model of perceptual encoding as estimation of the population parameter of a binominal distribution by Bayseian inference. We got a framework for theoretical analysis of the process from information acquisition to goods evaluation, by which we can analise the influence of goods'character and uncertainty of the law of cause and effect on goods evaluation at information processing level.
Physique indices have been used as convenient measures to assess body composition for school students. However, the measures could not accurately predict the percentage of Body Fat(%BF). This paper proposes a convenient, useful equation to estimate the%BF by using physique indices, physical fitness tests and motor ability tests that heve been widely used in schools. Physique, physical fitness and motor ability were measured for 613 junior high school students, and the%BF was estimated by Bioelectrical Impedance Analysis. Considering practicability, we selected the linear prediction model by which the%BF might be predicted directly without computing the body density. By all combination approach to variable selection, alternative prediction equations are obtained, and the final prediction equation is selected from the viewpoints of accuracy, rationality, stability, cross validity and practicability. Consequently, body mass index(BMI)and 50 m dash are found to be the best predictors, and the following equations are recommended for use in practice: Boy: %BF=0.962×(BMI)+2.52×(50m dash, s)-20.1(R=.745, SEE=2.43%)Girl: %BF=1.76×(BMI)+2.72×(50m dash, s)-31.9(R=.801, SEE=3.50%)where BMI=(body weight, kg)/(height, cm)2×104 %BF can be predicted with a significantly higher accuracy by adding 50 m dash as an additional predictor than by a single predictor, BMI.