No abstract

In current practice, regional models are limited in their capability to analyze policies involving changes and improvements to airports (and their services) and ground access transportation. Typically, airports are treated only as employment centers or as special generators. Important and distinct features of air passenger travel affecting trip distribution and mode choice are rarely modeled explicitly. This paper presents the development of a joint airport and ground access mode choice model for the New York City metropolitan region based on an extensive survey of airport users. Unlike travel to and from most U.S. cities, air passengers flying to and from the New York region face a nontrivial choice of airports and ground access modes (including premium transit options). A nested logit model was formulated with airport choice at the upper level and ground access mode choice at the second level; however, a multinomial logit model was found to be statistically preferable. Results indicate that air passenger travel behavior is significantly different for business and nonbusiness travelers. Overall, willingness to pay for trips to and from the airport is much higher than for regular intracity trips. Average yield, access time, and access cost are the most important determinants of air passenger's choice; demographics and trip characteristics are also significant. The developed tool was used for a comprehensive study of airport development alternatives in the New York region and is seen as the platform for additional data development and model extensions for future studies of air passenger service planning in the New York megaregion.

A substantial portion of regional travel is implemented by household members who travel together, primarily to participate in a shared household activity. Joint household travel is not explicitly accounted for in most regional travel models in which the unit of travel (either trip or tour) is considered for each person separately at each modeling stage—generation, mode, destination, and time-of-day choice. In addition, statistical evidence demonstrates that the vast majority of shared-ride travel consists of joint household travel. A modeling approach that distinguishes shared activity-based joint household travel from arranged interhouse-hold carpooling is clearly desirable to support accurate forecasts of shared-ride travel, critical in the evaluation of high-occupancy vehicle lanes or the adoption of toll strategies differentiated by occupancy levels. A range of aspects of joint travel both with empirical evidence and with discussion of modeling issues are addressed. A set of joint travel models is presented that has been estimated with the mid-Ohio regional travel household-interview survey. The model reported is one of the innovative components of the tour-based travel demand modeling system that has been developed for the Mid-Ohio Regional Planning Commission.

Existing approaches to modeling daily activity-travel patterns (DAPs) are mostly person based. However, intrahousehold interactions have a strong effect on the formation of the daily activity agenda of each household member. An approach is adopted that is based on the sequential modeling of the choices reflected in the DAPs of all household members, in a predetermined order of processing by person type, with explicit linkages between the choices made by different members of the household. The statistical analysis of intrahousehold interactions is described, as well as the application experience with this modeling approach in the framework of the new tour-based regional travel demand model recently developed for the Mid-Ohio Regional Planning Commission. Linkages across different household members that reflect the sharing of the same activities or the making of joint travel proved to be extremely strong statistically.

Microsimulation is increasingly assuming a major role in the advancement of demand-modeling practice. At the same time, it is attracting growing attention from the larger transportation-planning community. Four basic advantages of microsimulation versus conventional fractional-probability models are examined. The first is the technical advantage related to computational savings in the calculation and storage of large multidimensional probability arrays. The second is the meaningful advantage gained in the explicit modeling of various decision-making chains and time–space constraints on individual travel that allows for behavioral realism in the demand-modeling procedure. The third relates to the variability of microsimulation outcomes, which can yield full information about the distributions of the travel demand statistics of interest rather than single deterministic estimates or average values. As soon as constraints are introduced into the modeling framework (which often is done at the destination choice stage), competition arises, although generally it has been ignored in standard models. Microsimulation has the potential to handle this competition over work attractions and other travel activities in a meaningful fashion, which is the fourth advantage. These four advantages of microsimulation are discussed in light of the recent development and application of the New York best practice model, a microsimulation demand-modeling system for the New York–New Jersey–Connecticut metropolitan area.

With the ongoing debates from Florida to California and throughout the country concerning the benefits of high-speed rail, there is a renewed interest in intercity mode choice modeling. The investments for improving long-distance travel are substantial and may have serious impacts on travel demand, the environment and the economy. As such, alternatives for improving longdistance travel require careful evaluation before decisions are made on the form and design of long-distance travel infrastructure. A new nested multinomial logit mode-choice model has been developed that is sensitive to travel costs, distance, transit station accessibility, service frequency, number of transfers and parking costs. On the auto side the model considers the modes drive-alone and shared-ride with 2, 3 and 4 or more passengers. The transit side models regional bus, rail and air as modal options. To explore the model sensitivities, scenarios on increased gasoline prices and improved bus service are described. After a short introduction, the state-of-the-art of mode choice modeling is reviewed. 44 Section 3 explains how total travel demand is generated, and section 4 describes the mode choice model developed in this paper. Section 5 describes the application to the North Carolina Statewide Transportation Model (NCSTM) and section 6 shows the scenario application. The paper ends with conclusions and future1. INTRODUCTION In the last few years, a new interest in mode choice analysis has risen due to the controversy regarding the implementation of high-speed rail in different parts of the U.S. Analysis tools, however, have not caught up with this new demand in transportation modeling. The vast majority of mode choice models developed over the last few decades have been implemented for urban models with a focus on short-distance travel, where modal availability is different from longdistance travel. The travel behavior in long-distance travel is quite different, too, as people tend to be more familiar with modal options for short-distance travel than for long-distance travel. In addition, the composition of travelers differs. While short-distance use of transit is dominated by commuters, long-distance transit modes (particularly rail and air) are heavily used for pleasure trips as well as by business travelers. Given the fact that long-distance travelers tend to stay longer at their destination, travel time tends to be a less dominant factor in mode choice than in short-distance travel. Investments for improving long-distance travel often are tremendous. Adding a lane to an existing highway or even building a new highway may cost millions of dollars, just as adding a new rail line or improving the speed on an existing rail line may be cost-intensive. Environmental impacts may be serious, as increased auto traffic or air travel may increase gaseous emissions and noise levels substantially. Finally, the economic impact may be significant as well. According to Krugman [1], more accessible regions are ceteris paribus economically more successfu...

Household maintenance activities and associated tours, even if individually implemented, should be thought of as satisfying the needs of the entire household rather than simply the needs of the person who implements the activity. These activities are characterized by a high degree of substitution between household members who could implement them. This characteristic suggests a modeling structure in which maintenance activities are generated by the entire household and then allocated to the household members for implementation. In this approach, the modeling of allocated activities is executed by a sequence of two linked discrete choice models. The first model relates to the entire household and returns total daily frequency of individual maintenance tours by purpose. The second model relates to the tour level and returns the allocation of each generated tour to a particular household member. Within the general framework of a regional demand modeling system, these models are applied after, and are conditional on, the mandatory and joint tour generation stage. The statistical analysis and model estimation for this implementation of allocated household maintenance activities allow for important insights into the intrahousehold decision-making mechanism and improved travel demand forecasts. Models of this type have been implemented as components of the regional travel demand modeling system recently developed for the Mid-Ohio Regional Planning Commission.

This paper analyzes consumer choices over lunchtime restaurants using data from a sample of several thousand anonymous mobile phone users in the San Francisco Bay Area. The data is used to identify users approximate typical morning location, as well as their choices of lunchtime restaurants. We build a model where restaurants have latent characteristics (whose distribution may depend on restaurant observables, such as star ratings, food category, and price range), each user has preferences for these latent characteristics, and these preferences are heterogeneous across users. Similarly, each item has latent characteristics that describe users willingness to travel to the restaurant, and each user has individual-specific preferences for those latent characteristics. Thus, both users willingness to travel and their base utility for each restaurant vary across user-restaurant pairs. We use a Bayesian approach to estimation. To make the estimation computationally feasible, we rely on variational inference to approximate the posterior distribution, as well as stochastic gradient descent as a computational approach. Our model performs better than more standard competing models such as multinomial logit and nested logit models, in part due to the personalization of the estimates. We analyze how consumers re-allocate their demand after a restaurant closes to nearby restaurants versus more distant restaurants with similar characteristics, and we compare our predictions to actual outcomes. Finally, we show how the model can be used to analyze counterfactual questions such as what type of restaurant would attract the most consumers in a given location.