Estimation of a constrained multinomial logit model

Identifying the set of available alternatives in a choice process after considering an individual’s bounds or thresholds is a complex process that, in practice, is commonly simplified by assuming exogenous rules in the choice set formation. The Constrained Multinomial Logit (CMNL) model incorporates thresholds in several attributes as a key endogenous process to define the alternatives choice/rejection mechanism. The model allows for the inclusion of multiple constraints and has a closed form. In this paper, we study the estimation of the CMNL model using the maximum likelihood function, develop a methodology to estimate the model overcoming identification problems by an endogenous partition of the sample, and test the model estimation with both synthetic and real data. The CMNL model appears to be suitable for general applications as it presents a significantly better fit than the MNL model under constrained behaviour and replicates the MNL estimates in the unconstrained case. Using mode choice real data, we found significant differences in the values of times and elasticities between compensatory MNL and semi-compensatory CMNL models, which increase as the thresholds on attributes become active.     

Boundedly rational route choice behavior: A review of models and methodologies

Perfect rationality (PR) has been widely used in modeling travel behavior. As opposed to PR, bounded rationality (BR) has recently regained researchers’ attention since it was first introduced into transportation science in the 1980s due to its power in more realistic travel behavior modeling and prediction. This paper provides a comprehensive survey on the models of BR route choice behavior, aiming to identify current research gaps and provide directions for future research. Despite a small but growing body of studies on employing bounded rationality principle, BR route choice behavior remains understudied due to the following reasons: (a) The existence of BR thresholds leads to mathematically intractable properties of equilibria; (b) BR parameters are usually latent and difficult to identify and estimate; and (c) BR is associated with human being’s cognitive process and is challenging to model. Accordingly, we will review how existing literature addresses the aforementioned challenges in substantive and procedural bounded rationality models. Substantive bounded rationality models focus on choice outcomes while procedural bounded rationality models focus on the empirical studies of choice processes. Bounded rationality models in each category can be further divided based on whether time dimension is included. Accordingly, static and dynamic traffic assignment are introduced in substantive bounded rationality while two-stage cognitive models and day-to-day learning models in procedural bounded rationality are discussed. The methodologies employed in substantive bounded rationality include game theory and interactive congestion game, while those in procedural bounded rationality mainly adopt random utility and non- or semi-compensatory models. A comparison of all existing methodologies are given and bounded rationality models’ scope and boundaries in terms of predictability, transferability, tractability, and scalability are discussed. Finally existing research gaps are presented and several promising future research directions are given.     

Modeling travel mode choice of young people with differentiated E-hailing ride services in Nanjing China

E-hailing ride service (ERS) has become increasingly popular globally and is changing the urban mobility landscape. There is insufficient research effort in understanding the impact of ERS on travel behavior, in particular among young people. This paper aims to start filling that research gap by first collecting mode choice preference data through a stated preference survey in City of Nanjing, China and then applying nested logit (NL) models and a series of post-estimation analysis to address a number of key research questions of mode choice behavior without and with ERS. Three ERS modes are considered in the Chinese context: DiDi Taxi (D-Taxi), DiDi Express (D-Express), and DiDi Premier (D-Premier), all provided by DiDi Chuxing, the dominant ERS service provider in China. The study finds that age makes little difference in mode choice preference when ERS is introduced between the two age groups considered (18–30 and 31–45). The study results also suggest that young travelers are naturally drawn to ERS for what it represents (a technology innovation) and its business (pricing) model. ERS appears to be a competitive alternative to the conventional modes especially when they are under performed. The study also finds that ERS will likely increase vehicle kilometers traveled (VKT) considerably, which will lead to increase in on-road vehicular emissions, unless some mechanism to switch users to ridesharing is in place.     

Using census aggregates to proxy for household characteristics: an application to vehicle ownership

Traditionally, researchers studying transportation choice have used data either acquired from household surveys or broad, region-wide aggregates. At the disaggregate level, researchers usually do not have access to important variables or observations. This study investigates the potential usefulness of a proxy approach to modeling discrete choice vehicle ownership: substituting narrow area-based aggregate proxies for missing micro-level explanatory variables by accessing large, publicly maintained datasets. We use data from the 2000 Bay Area Travel Survey (BATS) and the contemporaneous U.S. Census file to compare three models of vehicle ownership, drawing area-wide proxies from increasing levels of aggregation. The models with proxies are compared with a parallel model that uses only survey data. The results indicate that the proxy models are preferred in terms of model selection criteria, and predict vehicle ownership as well or better than the survey model. Parameter values produced by the proxy method effectively approximate those returned by household survey models in terms of coefficient sign and significance, particularly when the aggregate variables are representative of their household-level counterparts. The proxy model with the narrowest level of aggregation achieved the best fit, coefficient precision, and percentage of correct prediction.

Parameter transfer of common-metric attributes in choice analysis: implications for willingness to pay

There is a growing literature that promotes the presence of a mix of compensatory and semi-compensatory processing strategies in the way that individuals evaluate packages of attributes in real or hypothetical markets, and make choices. This paper proposes a specification for the utility form in a choice model to test if, given a pair of attributes with a common-metric (e.g., components of travel time or cost), the attribute with the dominating level defines the marginal (dis)utility that is assigned to both attributes. We refer to this processing strategy as a parameter transfer rule. We use a stated choice data set, in the context of car driving individuals choosing between tolled and non-tolled routes, to estimate a mixed logit model which incorporates the presence of the parameter transfer rule and the conventional fully compensatory rule, both existing up to a probability. We find that if this parameter transfer heuristic is part of the mix, the WTP is more than 30% higher, on average, than when only a fully compensatory rule is imposed. We also contrast the parameter transfer rule with other semi-compensatory heuristics which have been investigated in other papers, and show that the finding adds further support to the accumulating evidence that a semi-compensatory attribute processing rules tend to result in higher mean WTP estimates compared to the fully compensatory attribute processing rule.     

A topological route choice model for metro

This article presents a route choice model for public transit networks that incorporates variables related to network topology, complementing those found in traditional models based on service levels (travel time, cost, transfers, etc.) and users’ socioeconomic and demographic characteristics (income level, trip purpose, etc.). The topological variables represent concepts such as the directness of the chosen route and user knowledge of the network. For both of these factors, the necessary data is endogenous to the modelling process and can be quantified without the need for information-gathering beyond what is normally required for building route choice models. Other novel variables in the proposed formulation capture notions of user comfort such as vehicle occupancy rates and certain physical characteristics of network stations. We conclude that these new variables significantly improve the explanatory and predictive ability of existing route choice specifications.     

Spatial transferability assessment of a composite walkability index: The Pedestrian Index of the Environment (PIE)

This paper analyzes the transferability of a composite walkability index, the Pedestrian Index of the Environment (PIE), to the Greater Montréal Area (GMA). The PIE was developed in Portland, Oregon, and is based on proprietary data. It combines six urban form variables into a score ranging from 20 to 100. The measure introduces several methodological refinements which have not been applied concurrently in previous efforts: a wide coverage of the different dimensions of the urban form, together with the use of a distance-based decay function and modelling-based weighing of the variables.This measure is applied to the GMA using local data in order to evaluate the feasibility of its transfer (the possibility of locally replicating the measure). It is then included in a series of mode choice models to assess its transferability (the capacity of the measure to describe walkability and predict mode choice in another urban area). The models, segmented by trip distance or trip purpose, are estimated and validated against observed trip data from the 2013 Origin-Destination survey.Significant positive correlation is found between the PIE and the choice of walking for short trips, for all purposes as well as for four specific trip purposes. The inclusion of the PIE also improves the accuracy of the modelling process as well as the prediction of the choice of walking for short trips. The PIE can therefore be used in the GMA, and potentially in other metropolitan areas, to improve the modelling of travel behavior for short trips.     

Modelling mode choice for freight transport using advanced choice experiments

In this paper we use advanced choice modelling techniques to analyse demand for freight transport in a context of modal choice. To this end, a stated preference (SP) survey was conducted in order to estimate freight shipper preferences for the main attributes that define the service offered by the different transport modes. From a methodological point of view, we focus on two critical issues in the construction of efficient choice experiments. Firstly, in obtaining good quality prior information about the parameters; and secondly, in the improved quality of the experimental data by tailoring a specific efficient design for every respondent in the sample.With these data, different mixed logit models incorporating panel correlation effects and accounting for systematic and random taste heterogeneity are estimated. For the best model specification we obtain the willingness to pay for improving the level of service and the elasticity of the choice probabilities for the different attributes. Our model provide interesting results that can be used to analyse the potential diversion of traffic from road (the current option) to alternative modes, rail or maritime, as well as to help in the obtaining of the modal distribution of commercial traffic between Spain and the European Union, currently passing through the Pyrenees.     

Combining web and face-to-face in travel surveys: comparability challenges?

Response rates for household travel surveys are tending to fall, and it seems unlikely that this trend will be reversed in the future. In recent years, travel data collection methods have evolved in order to obtain reliable data that are sufficiently detailed to feed increasingly complex models, and in order to integrate new technologies into survey protocols (Internet, GPS??). Combining different media is an obvious low-cost way of improving data quality as it increases the overall response rate. But the question of the comparability of data over time and between different survey modes remains unresolved. This paper makes a comparative analysis between the travel behaviours of web-based survey respondents and respondents to a face-to-face interview. The data were obtained from the 2006 Lyon conurbation household travel survey. Our analysis shows that the Internet respondents reported fewer trips per day than the face-to-face respondents (3.00 vs. 4.04 daily trips), and that the differences between the two groups varied according to the travel mode and trip purpose. While part of this difference can be explained by socioeconomic disparities (the Internet respondents had a specific profile) we cannot exclude the possibility of under-reporting due to the web medium.     

Development of railway station choice models to improve the representation of station catchments in rail demand models

ABSTRACT

This paper describes the development of railway station choice models suitable for defining probabilistic station catchments. These catchments can then be incorporated into the aggregate demand models typically used to forecast demand for new rail stations. Revealed preference passenger survey data obtained from the Welsh and Scottish Governments was used for model calibration. Techniques were developed to identify trip origins and destinations from incomplete address information and to automatically validate reported trips. A bespoke trip planner was used to derive mode-specific station access variables and train leg measures. The results from a number of multinomial logit and random parameter (mixed) logit models are presented and their predictive performance assessed. The models were found to have substantially superior predictive accuracy compared to the base model (which assumes the nearest station has a probability of one), indicating that their incorporation into passenger demand forecasting methods has the potential to significantly improve model predictive performance.     

A review of separate and joint strategies for the use of data on revealed and stated choices

This paper examines recent developments in household travel surveys that collect data for use in transportation planning and modeling efforts. The paper then introduces the notion of a total survey design and elaborates on what is meant by this concept. This is done first in the context of standard surveys of revealed choices. The paper discusses aspects of content and respondent burden and illustrates the potential to reduce respondent burden through careful consideration of content, question design, and question ordering. The paper also explores some issues of survey "friendliness" particularly with respect to activity surveys versus time-use surveys, with some observations about the potential of time-use surveys to eliminate some of the burden and content problems of previous diary designs. The remainder of the paper concentrates on the issue of collecting stated-response data and examines two alternative methods for collection: simultaneous collection of the contextual information and "on-the-fly" development of the alternatives for the stated-response questions, or sequential collection of contextual data and development of the stated-response questions. The paper also addresses issues of respondent burden that arise in the administration of stated-response surveys. The paper concludes with exploration of some of the reasons for collecting stated-response data, with particular emphasis on the US situation. In conclusion, the paper stresses again the need for a total design concept for collection of stated-response data, as well as for the simpler collection of more standard revealed choice data.     

Stated choice: a study in predictive validity using an aggregate truth set

Stated Choice models expand the ability of transportation planners to forecast future trends. The Stated Choice approach can forecast demand for new services or policies. However, Stated Choice models are subject to a range of experimental error not found within Revealed Preference (RP) designs. Primary among the concerns facing researchers is the ability of respondents to understand and operate upon hypothetical choice scenarios in a manner that will reproduce choices made under actual situations. These concerns are specified in the magnitude of a scaling factor. Efforts to estimate the scaling factor has proceeded by linking real decisions taken from a revealed preference survey with comparable decisions made under hypothetical conditions. However, where the alternative is new, actual decision data is not available. This study examines the level of error incorporated in a study where no RP data is available. The test of predictive validity focuses on the switching behavior of commuters at a single employment site. The actual data used to test the forecast is limited to company wide or aggregate ridership levels on the public transit mode taken two years after estimation of the SC model. The Fowkes and Preston hypothesis is examined and shown to bound the future actual value between forecasts derived from probabilistic and deterministic methods. The results show that with the passage of time, the probabilistic method approaches the reported ridership levels within 15 percent error.     

A fuzzy logic-genetic algorithm approach to modelling public transport users’ risk-taking behaviour

This paper seeks to determine the effects of uncertainty in out-of-vehicle times on route choice. Data were collected at two key interchanges in Auckland, New Zealand. Previous work modelled the data using a manual approach to fuzzy logic. This study extends that work by automating the process through defining a black-box function to match the survey data, then employing a genetic algorithm to fine-tune the fuzzy logic model. Results show that automation and the genetic algorithm improve the model’s capability to more accurately predict ridership. The tuning of the membership functions is conducted twice, first using initial fuzzy rules and again after the fuzzy rules have been adjusted to reduce disparity between the output and survey data. The calibrated membership functions provided for operational (transfer waiting and walking time and delay) and physical attributes (safety and seat availability) can be used by practitioners to determine an estimated ridership.     

Travel mode choice: affected by objective or subjective determinants?

This contribution presents theoretical considerations concerning the connections between life situation, lifestyle, choice of residential location and travel behaviour, as well as empirical results of structural equation models. The analyses are based on data resulting from a survey in seven study areas in the region of Cologne. The results indicate that lifestyles influence mode choice, although slightly, even when life situation is controlled for. The influence of life situation on mode choice exceeds the influence of lifestyle. The influence that lifestyle, and in part also life situation, has on mode choice is primarily mediated by specific location attitudes and location decisions that influence mode choice, respectively. Here objective spatial conditions as well as subjective location attitudes are important.

Flexing service schedules: Assessing the potential for demand-adaptive hybrid transit via a stated preference approach

This paper assesses the demand for a flexible, demand-adaptive transit service, using the Chicago region as an example. We designed and implemented a stated-preference survey in order to (1) identify potential users of flexible transit, and (2) inform the service design of the flexible transit mode. Multinomial logit, mixed-logit, and panel mixed-logit choice models were estimated using the data obtained from the survey. The survey instrument employed a d_p-efficient design and the Google Maps API to capture precise origins and destinations in order to create realistic choice scenarios. The stated-preference experiments offered respondents a choice between traditional transit, car, and a hypothetical flexible transit mode. Wait time, access time, travel time, service frequency, cost, and number of transfers varied across the choice scenarios. The choice model results indicate mode-specific values of in-vehicle travel time ranging between $16.3 per hour (car) and $21.1 per hour (flexible transit). The estimated value of walking time to transit is $25.9 per hour. The estimated value of waiting time at one’s point of origin for a flexible transit vehicle is $11.3 per hour; this value is significantly lower than the disutility typically associated with waiting at a transit stop/station indicating that the ‘at-home’ pick-up option of flexible transit is a highly desirable feature. The choice model results also indicate that respondents who use active-transport modes or public transit for their current commute trip, or are bikeshare members, were significantly more likely to choose flexible and traditional transit than car commuters in the choice experiments. The implications of these and other relevant model results for the design and delivery of flexible, technology-enabled services are discussed.     

Modeling household activity–travel interactions as parallel constrained choices

The daily activity-travel patterns of individuals often include interactions with other household members, which we observe in the form of joint activity participation and shared rides. Explicit representation of joint activity patterns is a widespread deficiency in extant travel forecasting models and remains a relatively under-developed area of travel behavior research. In this paper, we identify several spatially defined tour patterns found in weekday household survey data that describe this form of interpersonal decision-making. Using pairs of household decision makers as our subjects, we develop a structural discrete choice model that predicts the separate, parallel choices of full-day tour patterns by both persons, subject to the higher level constraint imposed by their joint selection of one of several spatial interaction patterns, one of which may be no interaction. We apply this model to the household survey data, drawing inferences from the household and person attributes that prove to be significant predictors of pattern choices, such as commitment to work schedules, auto availability, commuting distance and the presence of children in the household. Parameterization of an importance function in the models shows that in making joint activity-travel decisions significantly greater emphasis is placed on the individual utilities of workers relative to non-workers and on the utilities of women in households with very young children. The model and methods are prototypes for tour-based travel forecasting systems that seek to represent the complex interaction between household members in an integrated model structure.     

Network impacts of a road capacity reduction: Empirical analysis and model predictions

In spite of their widespread use in policy design and evaluation, relatively little evidence has been reported on how well traffic equilibrium models predict real network impacts. Here we present what we believe to be the first paper that together analyses the explicit impacts on observed route choice of an actual network intervention and compares this with the before-and-after predictions of a network equilibrium model. The analysis is based on the findings of an empirical study of the travel time and route choice impacts of a road capacity reduction. Time-stamped, partial licence plates were recorded across a series of locations, over a period of days both with and without the capacity reduction, and the data were ‘matched’ between locations using special-purpose statistical methods. Hypothesis tests were used to identify statistically significant changes in travel times and route choice, between the periods of days with and without the capacity reduction. A traffic network equilibrium model was then independently applied to the same scenarios, and its predictions compared with the empirical findings. From a comparison of route choice patterns, a particularly influential spatial effect was revealed of the parameter specifying the relative values of distance and travel time assumed in the generalised cost equations. When this parameter was ‘fitted’ to the data without the capacity reduction, the network model broadly predicted the route choice impacts of the capacity reduction, but with other values it was seen to perform poorly. The paper concludes by discussing the wider practical and research implications of the study’s findings.     

The relationship between land use and intrazonal trip making behaviors: Evidence and implications

This paper addresses the relationship between land use, destination selection, and travel mode choice. Specifically, it focuses on intrazonal trips, a sub-category of trip making where both trip origin and trip destination are contained in the same geographic unit of analysis, using data from the 1994 Household Activity and Travel Diary Survey conducted by Portland Metro. Using multinomial logit and binary logistic models to measure travel mode choice and decision to internalize trips, the evidence supports the conclusions that (1) intrazonal trips characteristics suggest mode choice for these trips might be influenced by urban form, which in turn affects regional trip distribution; (2) there is a threshold effect in the ability of economic diversity/mixed use to alter travel behavior; and (3) greater emphasis to destinations within the area where an individual’s home is located needs to be given in trip distribution models.     

Fleeing from Hurricane Irma: Empirical Analysis of Evacuation Behavior Using Discrete Choice Theory

This paper analyzes the observed decision-making behavior of a sample of individuals impacted by Hurricane Irma in 2017 (n = 645) by applying advanced methods based in discrete choice theory. Our first contribution is identifying population segments with distinct behavior by constructing a latent class choice model for the choice whether to evacuate or not. We find two latent segments distinguished by demographics and risk perception that tend to be either evacuation-keen or evacuation-reluctant and respond differently to mandatory evacuation orders.Evacuees subsequently face a multi-dimensional choice composed of concurrent decisions of their departure day, departure time of day, destination, shelter type, transportation mode, and route. While these concurrent decisions are often analyzed in isolation, our second contribution is the development of a portfolio choice model (PCM), which captures decision-dimensional dependency (if present) without requiring choices to be correlated or sequential. A PCM reframes the choice set as a bundle of concurrent decision dimensions, allowing for flexible and simple parameter estimation. Estimated models reveal subtle yet intuitive relations, creating new policy implications based on dimensional variables, secondary interactions, demographics, and risk-perception variables. For example, we find joint preferences for early-nighttime evacuations (i.e., evacuations more than three days before landfall and between 6:00 pm and 5:59 am) and early-highway evacuations (i.e., evacuations more than three days before landfall and on a route composed of at least 50% highways). These results indicate that transportation agencies should have the capabilities and resources to manage significant nighttime traffic along highways well before hurricane landfall.