Nobuhiro Sanko - Academia.edu (original) (raw)

Papers by Nobuhiro Sanko

Transport Policy Studies' Review

Business insight : the journal for deeper insights into business, 2010

Transportation Research Board 94th Annual MeetingTransportation Research Board, 2015

Use of recent data is crucial for travel demand forecasting. There is a significant body of evide... more Use of recent data is crucial for travel demand forecasting. There is a significant body of evidence that forecasts by models with recent data outperform those by models with older data. Even when the number of observations from the recent time point is significantly smaller than that from the older time point, utilising the former can result in better forecasts. However, forecasts made by a small number of observations are likely to be problematic where forecast performance is on average worse or is distributed with larger variance. Therefore, there must be a trade-off between data newness and number of observations. An opportunity exists to examine this trade-off in a context of commuting mode choice behaviours by utilising repeated cross-sectional data collected in Nagoya, Japan. Models are estimated utilising different number of observations (ranging from 50 to 10000) obtained from different time points (1971, 1981, and 1991), and they are applied to forecast commuting behaviours of 2001. Bootstrapping is adopted to have insights with statistical meaning. One of results is the following. Compared with models with larger number of observations (in the range of 550–10000 observations) from 1971, models using 1981 data with 300–500 observations and those using 1991 data with 200–250 observations produced statistically significantly better forecasts.

Kobe University monograph series in social science research, 2023

Transportation Research Interdisciplinary Perspectives

Widely used for analysing travel behaviour, SP (Stated Preference) data analysis examines respond... more Widely used for analysing travel behaviour, SP (Stated Preference) data analysis examines respondent preferences under hypothetical conditions. However, no consensus has been achieved concerning the appropriate approaches for SP data analysis and researchers continue to use the process of trial and error. Two of the issues examined in this study are closely related to each other: the response formats of the SP experiment design and the related modelling framework. Regarding the first issue, although the choice format is currently the most frequently used for transport research, attempts have been made to apply the iterative choice format that is commonly used in the CVM (Contingent Valuation Method). One of the iterative choice formats most frequently used, although not for transport analysis, applies a family of double-bounded (DB) formats. Regarding the second issue, the modelling framework must suit the design of the experiment (including the response format), since unsuitable ma...

Forecasts by disaggregate travel demand models are often based on data from the most recent time ... more Forecasts by disaggregate travel demand models are often based on data from the most recent time point, even when cross-sectional data are available from multiple time points. However, this is not a good use of data. In his previous work, the author proposed a method for improving the forecasting performance of models by utilising not only the most recent dataset but also an older dataset. He assumes that the parameters are functions of time, which means that parameter values vary over time. He analysed journeys to work mode choice behaviours in Nagoya, Japan. Behaviours in 2001 are forecast using a model with only the most recent 1991 dataset and models that combine the 1971, 1981, and 1991 datasets. His models using data from three time points outperformed the model using only 1991 dataset. This paper extends the author's work by expressing parameters as functions of gross domestic product (GDP) per capita. Although functions of time ascribe all reasons of parameter changes t...

This paper investigates alternative strategies for modelling car trip generation in the context o... more This paper investigates alternative strategies for modelling car trip generation in the context of the developing world. The proposed models particularly aim to capture the potential simultaneity between car trip generation and car ownership decisions, account for potential endogeneity, and deal with possible data limitations in the application context. In this regard, ordered response probit models of car trip generation (with and without the car ownership as an explanatory variable), are compared against combined models where car ownership and car trip decisions are estimated sequentially and jointly. The models are estimated using household survey data from two East African cities, Nairobi, Kenya and Dar-es-Salaam, Tanzania. The model performances are evaluated by comparing the coefficient values, the goodness-of-fit measures and the transferability statistics. Results show that though it is possible to overcome the particular modelling challenges using appropriate model structur...

We investigated household cars and motorcycles ownership behaviours in Asian big cities: Bangkok,... more We investigated household cars and motorcycles ownership behaviours in Asian big cities: Bangkok, Kuala Lumpur, Manila and Nagoya. The behaviours are modelled using bivariate ordered probit models and the impact of accessibility obtained from mode choice models on vehicle ownership is carefully examined. Comparison and temporal and spatial transferability analysis have brought insights into the accessibility measures useful for the modelling.

Transport Policy, 2013

ABSTRACT

IATSS Research, 2013

Disaggregate behaviour choice models have been improved in many aspects, but they are rarely eval... more Disaggregate behaviour choice models have been improved in many aspects, but they are rarely evaluated from the viewpoint of their ability to express intention to change travel behaviour. This study compared various models, including objective and latent models and compensatory and non-compensatory decision-making models. Latent models contain latent factors calculated using the LISREL (linear structural relations) model. Non-compensatory models are based on a lexicographic-semiorder heuristic. This paper proposes 'probability increment' and 'joint probability increment' as indicators for evaluating the ability of these models to express intention to change travel behaviour. The application to commuting travel data in the Chukyo metropolitan area in Japan showed that the appropriate non-compensatory and latent models outperform other models.

Journal of Choice Modelling, 2013

SP attribute values are sometimes set X times or 1/X times the RP attribute values in the directi... more SP attribute values are sometimes set X times or 1/X times the RP attribute values in the direction that changes respondents' RP behaviour. In estimating RP/SP models, assumptions on error structures are required: (a) RP and SP have both a common error component and independent error components (general model); (b) RP and SP have a common error component and SP only has an independent error component (SP-off-RP model); (c) RP and SP have independent error components only (independent model); and (d) RP and SP have a common error component only (double-bound model). This study simulates and examines the estimation efficiency of RP/SP models based on the D-error considering both error structures and attribute differences. Insights obtained are the following. (1) The general model offers better estimation efficiency in the neighbourhood of X ¼1.0. (2) For the SP-off-RP model and the independent model, the larger the variance in the error components of the SP model relative to the RP model, the larger the value of X that is required to minimise the D-error. The authors propose a method for designing an SP experiment in which the level-of-service of the SP differs from that of the RP by only a single attribute value.

While not generally discussed in applications, it is clear to most choice modellers that explanat... more While not generally discussed in applications, it is clear to most choice modellers that explanatory variables relating to alternatives and socio-demographic variables relating to decision makers are potentially affected by a number of important problems. These include measurement error, correlation with other unobserved factors, systematic errors such as lying or cheating by respondents, and missing values. Standard approaches to these problems are: to ignore the first two, to assume that the third does not happen and either to remove responses with the fourth problem or to impute missing values from observations of other respondents. This paper aims to propose a methodology to deal with these four problems through a latent variable approach. The variable chosen for the analysis is respondent income, a key variable in many applications of choice models, used to explain both heterogeneity in cost sensitivity and differences in underlying preferences for specific products or services...

Income is recognized as a key variable influencing consumer behavior, most notably in terms of a ... more Income is recognized as a key variable influencing consumer behavior, most notably in terms of a strong link with cost sensitivity. This is supported by a large body of empirical evidence using choice modeling techniques. Unfortunately, income information as available in most datasets used for modeling is affected by a number of issues; (1) it is measured as a categorical rather than as a continuous variable, (2) many respondents fail to provide income information, (3) the reported income is potentially correlated with other unobserved factors, and (4) there is scope for deliberate under or overstating of income by respondents. In the present paper, the authors propose to deal with these issues by replacing reported income with a latent income variable in the choice models, which at the same time is also used to explain the stated income in a measurement model. The proposed structure has a number of theoretical advantages. In comparison with using stated income, the auhtors should deal with some of the measurement error and bias issues. In comparison with using imputation of missing values, our model draws not just on data on stated income for those respondents without missing information, but the simultaneous estimation with the choice model means that the observed choices also inform the latent income variable. Two empirical applications using stated and revealed preference data illustrate the good performance of the method in practice.

Journal of Choice Modelling, 2014

Income is a key variable in many choice models. It is also one of the most salient examples of a ... more Income is a key variable in many choice models. It is also one of the most salient examples of a variable affected by data problems. Issues with income arise as measurement errors in categorically captured income, correlation between stated income and unobserved variables, systematic over-or understatement of income and missing income values for those who refuse to answer or do not know their (household) income. A common approach for dealing especially with missing income is to use imputation based on the relationship among those who report income between their stated income for reporters and their socio-demographic characteristics. A number of authors have also recently put forward a latent variable treatment of the issue, which has theoretical advantages over imputation, not least by drawing not just on data on stated income for reporters, but also choice behaviour of all respondents. We contrast this approach empirically with imputation as well as simpler approaches in two case studies, one with stated preference data and one with revealed preference data. Our findings suggest that, at least with the data at hand, the latent variable approach produces similar results to imputation, possibly an indication of non-reporters of income having similar income distributions from those who report it. But in other data sets the efficiency advantage over imputation could help in revealing issues in the complete and accurate reporting of income.