Transportation sector has been a major contributing factor to the overall emissions of most pollutants and thus their impacts on the environment. Among all transportation activities, on-road travel accounts for most part of the Greenhouse gas (GHG) emissions and fuel use. It also has a very un-desirable impact on the transportation network conditions increasing the traffic congestion levels. The main aim of transportation planning agencies is to implement the policy changes that will reduce automobile dependency and increase transit and non-motorized modes usage. However, planning agencies can come up with proactive economic, land-use and transportation policies provided they have a model which is sensitive to all the above mentioned factors to predict the vehicle fleet composition and usage of households. Moreover, the type of vehicle that a household gets (vehicle type choice) and the annual mileage (usage) associated with that vehicle is very closely related to the person in the household who uses that vehicle the most (allocation to primary driver). So, it is no longer possible to view all these decisions separately. Instead, we need to model all these decisions- vehicle type choice, usage, and allocation to primary driver simultaneously at a household level. In this study, we estimate and apply a joint household-level model of the number of vehicles owned by the household, the vehicle type choice of each vehicle, the annual mileage on each vehicle, as well as the individual assigned as the primary driver for each vehicle. A version of the proposed model system currently serves as the engine for a household vehicle composition and evolution simulator, which itself has been embedded within the larger SimAGENT (for Simulator of Activities, Greenhouse emissions, Networks, and Travel) activity-based travel and emissions forecasting system for the Southern California Association of Governments (SCAG) planning region. / text

This thesis compares different approaches to estimating budgets for Kuhn-Tucker (KT) demand systems, more specifically for the multiple discrete-continuous extreme value (MDCEV) model. The approaches tested include: (1) The log-linear regression approach (2) The stochastic frontier regression approach, and (3) arbitrarily assumed budgets that are not necessarily modeled as a function of decision maker characteristics and choice-environment characteristics. The log-linear regression approach has been used in the literature to model the observed total expenditure as way of estimating budgets for the MDCEV models. This approach allows the total expenditure to depend on the characteristics of the choice-maker and the choice environment. However, this approach does not offer an easy way to allow the total expenditure to change due to changes in choice alternative-specific attributes, but only allows a reallocation of the observed total expenditure among the different choice alternatives. To address this issue, we propose the stochastic frontier regression approach. The approach is useful when the underlying budgets driving a choice situation are unobserved, but only the expenditures on the choice alternatives of interest are observed. The approach is based on the notion that consumers operate under latent budgets that can be conceived (and modeled using stochastic frontier regression) as the maximum possible expenditure they are willing to incur. To compare the efficacy of the above-mentioned approaches, we performed two empirical assessments: (1) The analysis of out-of-home activity participation and time-use (with a budget on the total time available for out-of-home activities) for a sample of non-working adults in Florida, and (2) The analysis of household vehicle type/vintage holdings and usage (with a budget on the total annual mileage) for a sample of households in Florida. A comparison of the MDCEV model predictions (based on budgets from the above mentioned approaches) demonstrates that the log-linear regression approach and the stochastic frontier approach performed better than arbitrarily assumed budgets approaches. This is because both approaches consider heterogeneity in budgets due to socio-demographics and other explanatory factors rather than arbitrarily imposing uniform budgets on all consumers. Between the log-linear regression and the stochastic frontier regression approaches, the log-linear regression approach resulted in better predictions (vis-à-vis the observed distributions of the discrete-continuous choices) from the MDCEV model. However, policy simulations suggest that the stochastic frontier approach allows the total expenditures to either increase or decrease as a result of changes in alternative-specific attributes. While the log-linear regression approach allows the total expenditures to change as a result of changes in relevant socio-demographic and choice-environment characteristics, it does not allow the total expenditures to change as a result of changes in alternative-specific attributes.

Spatial transferability of travel forecasting models, or the ability to transfer models from one geographical region to another, can potentially help in significant cost and time savings for regions that cannot invest in extensive data-collection and model-development procedures. This issue is particularly important in the context of tour-based/activity-based models whose development typically involves significant data inputs, skilled staff, and long production times. However, most literature on model transferability has been in the context of traditionally used trip-based models, particularly for linear regression-based trip generation and logit-based mode choice models, with little evidence on the transferability of activity-based models and that of emerging model structures. The overarching goal of this dissertation is to assess the spatial transferability of activity-based travel demand models. To this end, the specific objectives are to: 1. Survey the literature to synthesize: (a) the approaches used to transfer models, (b) the metrics used to assess model transferability, (c) the available evidence on spatial transferability of travel models, and (d) notable gaps in literature; 2. Lay out a framework for assessing the spatial transferability of activity-based travel forecasting model systems, and evaluate alternative methods/metrics used for assessing the transferability of specific model components and their parameters; 3. Conduct empirical assessments of spatial transferability of the following two model components used in today's activity-based model systems: (a) daily activity participation and time-use models, and (b) tour-based time-of-day choice models. Data from the 2009 National Household Travel Survey (NHTS) and the 2000 San Francisco Bay Area Travel Survey (BATS) were used for these empirical assessments; 4. Conduct empirical assessments of model transferability using emerging model structures that have begun to be used in activity-based model systems - specifically the multiple discrete-continuous extreme value (MDCEV) model; 5. Investigate alternate ways of enhancing model transferability; specifically: (a) pooling data from different geographical regions, and (b) improvements to the model structure. The dissertation provides a framework for assessing the transferability of activity-based models systems, along with empirical evidence on the pros and cons of alternative methods and metrics of transferability assessment. The results suggest the need to consider model sensitivity to changes in explanatory variables as opposed to relying solely on the ability to predict aggregate distributions. Updating the constants of a transferred model using local data (a widely used method to transfer models) was found to help in increasing the model's ability to predict aggregate patterns but not necessarily in enhancing its sensitivity to changes in explanatory variables. Also, transferability assessments ought to consider sampling variance in parameter estimates as opposed to only the point estimates. Empirical analysis with the daily activity participation and time-use model shed new light on the prediction properties of the MDCEV model structure that have implications for model transferability. This led to the development of a new model structure called the multiple discrete continuous heteroscedastic extreme value (MDCHEV) model that incorporates heteroscedasticity in the model's stochastic distributions and helps in enhancing model transferability. Transferability assessment of the time-of-day choice models show encouraging evidence of transferability of a large proportion of the model coefficients, albeit except important parameters such as the travel time coefficients. Collectively, there is evidence that pooling data from multiple regions may help in building better transferable models than those transferred from a single region.

The United States (US) witnessed remarkable growth in annual long distance travel over the past few decades. Over half of the long distance travel in the US is made for pleasure, including visiting friends and relatives (VFR) and leisure activities. This trend could continue with increased use of information and communication technologies for socialization, and enhanced mobility being achieved using fuel-efficient (electric/hybrid) and technology enhanced vehicles. Despite these developments, and recent interest to implement alternate mass transit options to serve this market, not much exists on the measurement, analysis and modeling of long distance pleasure travel in the U.S. Statewide and national models are used to estimate long distance travel, but these are predominantly trip-based models, making it difficult to understand long distance trips as collection of household-level travel behavior. This form of travel behavior has been studied a lot in tourism, but in a piecemeal manner, such as to (from) a specific destination. Further, most of these studies are confined to analyzing leisure market, with VFR market gaining recognition only recently. In essence, annual household long distance pleasure travel behavior needs to be studied in a comprehensive manner rather than as isolated trips. This is because, most of these household travel decisions are undertaken considering their annual time and monetary budget, and their perceived cost to travel to one (or more) destination for given pleasure purpose on one (or more) occasion using a given mode of travel. Thus, the main objective of this dissertation is to develop a comprehensive behavioral model framework to analyze the above-discussed annual household long distance pleasure travel choices. To start the above effort, it is first required to collect detailed annual household travel data, last collected over two decades ago (e.g.: ATS, 1995). No such recent effort has been pursued due to the significant labor and economic resource required to undertake it. There exist recent surveys (NHTS, 2001), but collected over a shorter (four week) period, and require significant processing even to arrive at aggregate annual travel estimates. Second, besides surveys, there is a need for additional data to estimate households’ annual pleasure travel budget, and their cost to travel and stay at each of their potential destination choices, which are not readily available. Thus, as the first goal, this dissertation analyzes long distance travel reported across historical surveys (NPTS; ATS; NHTS), to understand the differences in their definition, enumeration of purpose and collection methods. The intent here is twofold, first to conceive a method to estimate annual travel from surveys with shorter collection period. Further, the second intent is to gather travel patterns from these historical datasets such that it informs the second goal of this dissertation, i.e. development of a behavioral framework to analyze annual household pleasure travel. To this effect, this research also analyzes pleasure expenditures using Consumer Expenditure Survey (CEX, BLS) data. Interestingly, the analysis reveals CEX pleasure travel expenditure pattern to be similar to the travel pattern reported for the same market segments in travel survey (ATS). Importantly, the above analysis informs the development of behavioral models, pursued as two distinct tasks to achieve the second goal. As the first task, a novel econometric model and forecasting procedure is developed to analyze a household’s annual long distance leisure travel decisions. Specifically, a households’ time spent across one (or more) destination and travel mode to such destination for leisure is modeled subject to time and money budget constraints. In this methodological framework, the destination choice is modeled as a continuous variable (time at destination) using Multiple-Discrete Continuous Extreme Value model (MDCEV). While, travel mode choice to these destination(s) are modeled as a discrete choice, through a nested Multinomial Logit Model (MNL), with price variation introduced across the above choice of destination(s) and travel modes (air/ground). This required estimating annual monetary budgets, travel cost and per night lodging cost for each sample household, with each of them having 210 potential destinations and 2 travel mode choices respectively. The second task, involved the development of a broader national model system to analyze households’ annual pleasure travel decisions such as: choice (duration) at destination(s), travel purpose (VFR or leisure), mode (airplane or auto) choice and trip frequencies to these destination(s) using the same dataset. It was modeled in two stages, with the first stage estimating households’ annual pleasure time budget using a stochastic frontier model. This budget was then used as constraint to analyze households’ annual choice of destination and purpose using a nested MDCEV-MNL model in the second stage. A log sum variable from a nested joint multinomial logit model of trip frequency and mode choice for each purpose (VFR or leisure) is also introduced as input at this stage. This model was then validated using a prediction procedure, and further applied to test a policy scenario (increase in travel cost). The above national pleasure travel demand model could be further enhanced by including monetary constraints and price variation as in the first task. Overall, the model system proposed in this dissertation forms the foundation for a national comprehensive long distance travel model. This could be achieved through inclusion of other prominent travel purpose such as business and commuting to the national travel demand model presented in this research.

Spatial transferability of travel demand models has been an issue of considerable interest, particularly for small and medium sized planning areas that often do not have the resources and staff time to collect large scale travel survey data and estimate model components native to the region. With the advent of more sophisticated microsimulation-based activity-travel demand models, the interest in spatial transferability has surged in the recent past as smaller metropolitan planning organizations seek to take advantage of emerging modeling methods within the limited resources they can marshal. Traditional approaches to identifying geographical contexts that may borrow and transfer models between one another involve the exogenous a priori identification of a set of variables that are used to characterize the similarity between geographic regions. However, this ad hoc procedure presents considerable challenges as it is difficult to identify the most appropriate criteria a priori. To address this issue, this thesis proposes a latent segmentation approach whereby the most appropriate criteria for identifying areas with similar profiles are determined endogenously within the model estimation phase, customized for every model type. The end products are a set of optimal similarity measures that link regions to one another as well as a fully transferred model, segmented to account for heterogeneity in the population. The methodology is demonstrated and its efficacy established through a case study that utilizes the National Household Travel Survey (NHTS) dataset for information on weekday activities unemployed individuals within 9 regions in the states of California and Florida engage in. A multiple discrete continuous extreme value (MDCEV) model is developed that simulates the discrete nature of activity selection as well as the continuous nature of activity participation. The estimated model is then applied onto the Austin–San Marcos MSA, a context withheld from the original estimation in order to assess its performance. The performance of the segmented model was then examined vis-à-vis that of other models that are similar to the local region in only one dimension. It is found that the methodology offers a robust mechanism for identifying latent segments and establishing criteria for transferring models between areas. / text