This dissertation investigates the evacuees' behavior under hurricane evacuation conditions and develops an agent-based travel demand model system for hurricane evacuation simulation using these behavioral findings. The dissertation econometrically models several important evacuation decisions including evacuate-stay, accommodation type choice, evacuation destination choice, evacuation mode choice, departure time choice, and vehicle usage choice. In addition, it explicitly considers the pre-evacuation preparation activities using activity-based approach. The models are then integrated into a two-module agent-based travel demand model system. The dissertation first develops the evacuate-stay choice model using the random-coefficient binary logit specification. It uses heterogeneous mean of the random parameter across households to capture shadow evacuation. It is found that the likelihood of evacuation for households that do not receive any evacuation notice decreases as their distance to coast increase on average. The distance sensitivity factor, or DSF, is introduced to construct the different scenarios of geographical extent of shadow evacuation. The dissertation then conducts statistical analysis of the vehicle usage choice. It identifies the contributing factors to households' choice of the number of vehicles used for evacuation and develop predictive models of this choice that explicitly consider the constraint imposed by the number of vehicles owned by the household. This constraint is not accommodated by ordered response models. Data comes from a post-storm survey for Hurricane Ivan. The two models developed are variants of the regular Poisson regression model: the Poisson model with exposure and right-censored Poisson regression. The right-censored Poisson model is preferred due to its inherent capabilities, better fit to the data, and superior predictive power. The multivariable model and individual variable analyses are used to investigate seven hypotheses. Households traveling longer distances or evacuating later are more likely to use fewer vehicles. Households with prior hurricane experience, greater numbers of household members between 18 and 80, and pet owners are more likely to use a greater number of vehicles. Income and distance from the coast are insignificant in the multivariable models, although their individual effects have statistically significant linear relationship. However, the Poisson based models are non-linear. The method for using the right-censored Poisson model for producing the desired share of vehicle usage is also provided for the purpose of generating individual predictions for simulation. The dissertation then presents a descriptive analysis of and econometric models for households' pre-evacuation activities based on behavioral intention data collected for Miami Beach, Florida. The descriptive analysis shows that shopping - particularly food, gasoline, medicine, and cash withdrawal - accounts for the majority of preparation activities, highlighting the importance of maintaining a supply of these items. More than 90% of the tours are conducted by driving, emphasizing the need to incorporate pre-evacuation activity travel into simulation studies. Households perform their preparation activities early in a temporally concentrated manner and generally make the tours during daylight. Households with college graduates, larger households, and households who drive their own vehicles are more likely to engage in activities that require travel. The number of household members older than 64 has a negative impact upon engaging in out-of-home activities. An action day choice model for the first tour suggests that households are more likely to buy medicine early but are more likely to pick up friends/relatives late. Households evacuating late are more likely to conduct their activities late. Households with multiple tours tend to make their first tour early. About 10% of households chain their single activity chains with their ultimate evacuation trips. The outcomes of this paper can be used in demand generation for traffic simulations. The dissertation finally uses the behavioral findings and develops an agent-based travel demand model system for hurricane evacuation simulation, which is capable of generating the comprehensive household activity-travel plans. The system implements econometric and statistical models that represent travel and decision-making behavior throughout the evacuation process. The system considers six typical evacuation decisions: evacuate-stay, accommodation type choice, evacuation destination choice, mode choice, vehicle usage choice and departure time choice. It explicitly captures the shadow evacuation population. In addition, the model system captures the pre-evacuation preparation activities using an activity-based approach. A demonstration study that predicts activity-travel patterns using model parameters estimated for the Miami-Dade area is discussed. The simulation results clearly indicate the model system produced the distribution of choice patterns that is consistent with sample observations and existing literature. The model system also identifies the proportion of the shadow evacuation population and their geographical extent. About 23% of the population outside the designated evacuation zone would evacuate. The shadow evacuation demand is mainly located within 3.1 miles (5 km) of the coastline. The output demand of the model system works with agent-based traffic simulation tools and conventional trip-based simulation tools. The agent-based travel demand model system is capable of generating activity plans that works with agent-based traffic simulation tools and conventional trip-based simulation tools. It will facilitate the hurricane evacuation management. / Ph. D.

This thesis explores a border security system based on agent-based modeling and simulation (ABMS). The ABMS software platform, map aware non-uniform automata, is used to model various scenarios and evaluate the border security system given a set of infiltrators who have evolutionary behavior governed by a genetic algorithm (GA). The GA is used to represent adaptive behavior of the enemy when the friendly force has deployed our border security at a maximum level. By using a near optimal Latin hypercube design, our simulation runs are implemented efficiently and the border security system is analyzed using four different kinds of measures of effectiveness. / text

Traffic congestion has been placing an extremely high burden on the development of modern cities. Congestion can be alleviated by either increasing road capacity, or by reducing traffic demand. For decades, increasing capacity by building more roads and lanes has been the major solution applied to accommodate the ever-growing traffic demand. However, it turns out to be of limited effect due to some well-known phenomenon such as latent demand. Controlling and managing traffic demand has in turn been viewed as a cost-effective alternative to increasing road capacity, as has been demonstrated many successful applications all around the world. Within the concept framework of Traffic Demand Management (TDM), Active Transportation and Demand Management (ATDM) is the dynamic management, control, and influence of traffic demand and traffic flow of transportation facilities. ATDM strategies attempt to influence traveler behavior and further manage traffic flow in a time-dependent manner within the existing infrastructure Successful ATDM applications include congestion pricing, adaptive ramp metering, dynamic speed limits, dynamic lane use control, etc. Singapore stands out to be an excellent success story of ATDM, as the implementations of "Cap and Trade" license plates and electronic road pricing make motoring a high cost privilege for citizens of Singapore, making the public relies on transit. Monetary leverage is an effective instrument to facilitate ATDM. Examples of ATDM applications adopting monetary instrument includes dynamic congestion pricing, "Cap and Trade" of car licenses, etc. Taking congestion pricing as an example, policy makers are inducing travelers' behavior and alternating their preferences towards different behavior decisions by levying price tags to different choices. As an important underpinning of rationing choice theory, an individual assigns an ordinal number over the available actions and this ordinal number is calculated by their utility function or payoff function. The individual's preference is expressed as the relationship between those ordinal assignments. In the implementation of congestion pricing, policy makers are imposing an additional high disutility to congested roads and therefore pushing some of the travelers to take alternative routes or shift to alternative departure times or even cancel the trips. However, congestion pricing suffers from public aversion as it creates burden on the motoring of low-income people and therefore doesn't help to alleviate social inequality. The concept of Tradable Mobility Credit (TMC) has been proposed by a group of researchers as another innovative application to facilitate dynamic traffic demand management and solve social inequality issues using pricing instruments. The concept of TMC is borrowed from carbon trading in environmental control. A limited quota of personal auto usage is issued to eligible travelers and credits can be traded in a free market fashion. This guarantees that the roadway usage does not exceed capacity while avoiding the negative effects of shortages normally associated with quotation systems. TMC is literally not a market-ready policy as the integration of the supporting infrastructures, including the trading market, the credit assignment component, and the credit charging component, has not been fully explored yet. Existing TMC research focuses on explaining and exploring the equilibrium condition through analytical methods such as mathematical modeling. Analytical models produce perfect convergence curves and deterministic equilibrium traffic flow patterns. Analytical models provide influential guidance for further works but the solution procedure may encounter problems when dealing with larger real world networks and scenarios. Meantime, current analytical models don't consider the microstructure of the credit trading market sufficiently while it's actually the most unique component of TMC system. Motivated by those concerns, an integrated TMC evaluation platform consisting of a policy making module and traveler behavior modules are proposed in this research. The concept of Agent-Based Modeling and Simulation (ABMS) is extensively adopted in this integrated platform as each individual traveler carries his/her personal memory across iterations. The goal of establishing this framework is to better predict a traveler's route choice and trading behavior if TMC is imposed and further provide intelligence to potential policy makers' decision making process. The proposed integrated platform is able to generate results at different aggregation levels, including both individual level microscopic behavior data as well as aggregated traffic flow and market performance data. In order to calibrate the proposed integrated platform, an online interactive experiment is designed based on an experimental economic package and a human research element with 22 participants has been conducted on this experiment platform to gather field data regarding a real person's route choice behavior and credit trading behavior in an artificial TMC system. Participants are recruited from forum, listserve, social media, etc. The calibrated platform is proved to have the ability to predict travelers' behavior accurately. A prototype market microstructure is proposed in this research as well and it is proved to be a cost-effective setting and resulted to a vast amount of economic saving given the fact that travelers would behave similar to the prediction generated by traveler behavior module. It's also demonstrated that the principle of Pareto-improving is not achieved in the proposed ABMS models.

This thesis aims to develop an alternative expectations model to the Rational Expectations Hypothesis (REH) and adaptive-expectations models, which provides more accurate temporal predictive performance and more closely reflects recent advances in behavioural economics, the ‘science of complexity’ and network dynamics. The model the thesis develops is called Adaptive Interactive Expectations (AIE), a subjective dynamic model of the process of expectations formation. To REH, the AIE model provides both an alternative and a complement. AIE and REH complement one another in that they are diametrically opposite in the following five dimensions, agent intelligence, agent interaction, agent homogeneity, equilibrium assumptions and the rationalisation process. REH and AIE stress the importance of hyper-intelligent agents interacting only via a price signal and near zero-intelligent agents interacting via a network structure, respectively. The complementary nature of AIE and REH provide dual perspectives that enhance analysis. The Dun & Bradstreet (D&B 2008) profit expectations survey is used in the thesis to calibrate AIE and make predictions. The predictive power of the AIE and REH models is compared. The thesis introduces the ‘pressure to change profit expectations index’, px. This index provides the ability to model unknowns within an adaptive dynamic process and combine the beliefs from interactive-expectations, adaptive-expectations and biases that include pessimism, optimism and ambivalence. AIE uses networks to model the flow of interactive-expectations between firms. To overcome the uncertainty over the structure of the interactive network, the thesis uses model-averaging over 121 network topologies. These networks are defined by three variables regardless of their complexity. Unfortunately, the Bayesian technique’s use of the number of variables as a measure of complexity makes it unsuitable for model-averaging over the network topologies. To overcome this limitation in the Bayesian technique, the thesis introduces two model-averaging techniques, ‘runtime-weighted’ and ‘optimal-calibration’. These model-averaging techniques are benchmarked against ‘Bayes-factor model-averaging’ and ‘equal-weighted model-averaging’. In addition to the aggregate called all–firms, the D&B (2008) survey has four divisions, manufacturing durables, manufacturing non–durables, wholesale and retail. To make use of the four divisions, the thesis introduces a ‘link-intensity matrix’ based upon an ‘input-output table’ to improve the calibration of the networks. The transpose of the table is also used in the thesis. The two ‘link-intensity matrices’ are benchmarked against the default, a ‘matrix of ones’. The aggregated and disaggregated versions of AIE are benchmarked against adaptive-expectations to establish whether the interactive-expectations component of AIE add value to the model. The thesis finds that AIE has more predictive power than REH. ‘Optimal-calibration model-averaging’ improves the predictive performance of the better-fitting versions of AIE, which are those versions that use the ‘input-output table’ and ‘matrix of ones’ link-intensity matrices. The ‘runtime-weighted model-averaging’ improves the predictive performance of only the ‘input-output table’ version of AIE. The interactive component of the AIE model improves the predictive performance of all versions of the AIE over adaptive-expectations. There is an ambiguous effect on prediction performance from introducing the ‘input-output table’. However, there is a clear reduction in the predictive performance from introducing its transpose. AIE can inform the debate on government intervention by providing an Agent-Based Model (ABM) perspective on the conflicting mathematical and narrative views proposed by the Greenwald–Stiglitz Theorem and Austrian school, respectively. Additionally, AIE can provide a complementary role to REH, which is descriptive/predictive and normative, respectively. The AIE network calibration uses an ‘input-output table’ to determine the link-intensity; this method could provide Computable General Equilibrium (CGE) and Dynamic Stochastic General Equilibrium (DSGE) with a way to improve their transmission mechanism. Furthermore, the AIE network calibration and prediction methodology may help overcome the validation concerns of practitioners when they implement ABM.

This thesis aims to develop an alternative expectations model to the Rational Expectations Hypothesis (REH) and adaptive-expectations models, which provides more accurate temporal predictive performance and more closely reflects recent advances in behavioural economics, the ‘science of complexity’ and network dynamics. The model the thesis develops is called Adaptive Interactive Expectations (AIE), a subjective dynamic model of the process of expectations formation. To REH, the AIE model provides both an alternative and a complement. AIE and REH complement one another in that they are diametrically opposite in the following five dimensions, agent intelligence, agent interaction, agent homogeneity, equilibrium assumptions and the rationalisation process. REH and AIE stress the importance of hyper-intelligent agents interacting only via a price signal and near zero-intelligent agents interacting via a network structure, respectively. The complementary nature of AIE and REH provide dual perspectives that enhance analysis. The Dun & Bradstreet (D&B 2008) profit expectations survey is used in the thesis to calibrate AIE and make predictions. The predictive power of the AIE and REH models is compared. The thesis introduces the ‘pressure to change profit expectations index’, px. This index provides the ability to model unknowns within an adaptive dynamic process and combine the beliefs from interactive-expectations, adaptive-expectations and biases that include pessimism, optimism and ambivalence. AIE uses networks to model the flow of interactive-expectations between firms. To overcome the uncertainty over the structure of the interactive network, the thesis uses model-averaging over 121 network topologies. These networks are defined by three variables regardless of their complexity. Unfortunately, the Bayesian technique’s use of the number of variables as a measure of complexity makes it unsuitable for model-averaging over the network topologies. To overcome this limitation in the Bayesian technique, the thesis introduces two model-averaging techniques, ‘runtime-weighted’ and ‘optimal-calibration’. These model-averaging techniques are benchmarked against ‘Bayes-factor model-averaging’ and ‘equal-weighted model-averaging’. In addition to the aggregate called all–firms, the D&B (2008) survey has four divisions, manufacturing durables, manufacturing non–durables, wholesale and retail. To make use of the four divisions, the thesis introduces a ‘link-intensity matrix’ based upon an ‘input-output table’ to improve the calibration of the networks. The transpose of the table is also used in the thesis. The two ‘link-intensity matrices’ are benchmarked against the default, a ‘matrix of ones’. The aggregated and disaggregated versions of AIE are benchmarked against adaptive-expectations to establish whether the interactive-expectations component of AIE add value to the model. The thesis finds that AIE has more predictive power than REH. ‘Optimal-calibration model-averaging’ improves the predictive performance of the better-fitting versions of AIE, which are those versions that use the ‘input-output table’ and ‘matrix of ones’ link-intensity matrices. The ‘runtime-weighted model-averaging’ improves the predictive performance of only the ‘input-output table’ version of AIE. The interactive component of the AIE model improves the predictive performance of all versions of the AIE over adaptive-expectations. There is an ambiguous effect on prediction performance from introducing the ‘input-output table’. However, there is a clear reduction in the predictive performance from introducing its transpose. AIE can inform the debate on government intervention by providing an Agent-Based Model (ABM) perspective on the conflicting mathematical and narrative views proposed by the Greenwald–Stiglitz Theorem and Austrian school, respectively. Additionally, AIE can provide a complementary role to REH, which is descriptive/predictive and normative, respectively. The AIE network calibration uses an ‘input-output table’ to determine the link-intensity; this method could provide Computable General Equilibrium (CGE) and Dynamic Stochastic General Equilibrium (DSGE) with a way to improve their transmission mechanism. Furthermore, the AIE network calibration and prediction methodology may help overcome the validation concerns of practitioners when they implement ABM.

Supply chain management is a complex process requiring the coordination of numerous decisions in the attempt to balance often-conflicting objectives such as quality, cost, and on-time delivery. To meet these and other objectives, a focal company must develop organized systems for establishing and managing its supplier relationships. A reliable, decision-support tool is needed for selecting the best procurement strategy for each supplier, given knowledge of the existing sourcing environment. Supplier segmentation is a well-established and resource-efficient tool used to identify procurement strategies for groups of suppliers with similar characteristics. However, the existing methods of segmentation generally select strategies that optimize performance during normal operating conditions, and do not explicitly consider the effects of the chosen strategy on the supply chain’s ability to respond to disruption. As a supply chain expands in complexity and scale, its exposure to sources of major disruption like natural disasters, labor strikes, and changing government regulations also increases. With increased exposure to disruption, it becomes necessary for supply chains to build in resilience and robustness in the attempt to guard against these types of events. This work argues that the potential impacts of disruption should be considered during the establishment of day-to-day procurement strategy, and not solely in the development of posterior action plans. In this work, a case study of a laser printer supply chain is used as a context for studying the effects of different supplier segmentation methods. The system is examined using agent-based modeling and simulation with the objective of measuring disruption impact, given a set of initial conditions. Through insights gained in examination of the results, this work seeks to derive a set of improved rules for segmentation procedure whereby the best strategy for resilience and robustness for any supplier can be identified given a set of the observable supplier characteristics.

京都大学 / 0048 / 新制・課程博士 / 博士(経済学) / 甲第22717号 / 経博第620号 / 新制||経||294(附属図書館) / 京都大学大学院経済学研究科経済学専攻 / (主査)教授 関口 倫紀, 教授 若林 直樹, 教授 椙山 泰生 / 学位規則第4条第1項該当 / Doctor of Economics / Kyoto University / DGAM

The modeling of financial risk, whose shortcomings came to the fore during the financial crisis, generally understands risk from the history of prices and returns. However, the state space of risk is not fully revealed from the history of prices and returns. In this dissertation, certain cognitive biases were modeled, and the simulation results were quantitatively characterized to reveal risk not revealed from the history of prices and returns. This contribution adds to the extant literature on the modeling of financial risk by showing how to reveal parts of the state space of risk not revealed from other methods in use today.

Modeling and simulation of military operations are becoming popular with the widespread application of artificial intelligence methods. As the decision makers would like to analyze the results of the simulations in greater details, entity-level simulation of physical world and activities of actors (soldiers, tanks, etc) is unavoidable. In this thesis, a multiagent framework for simulating task driven autonomous activities of actors or group of actors is proposed. The framework is based on BDI-architecture where an agent is composed of beliefs, goals and plans. Besides, an agent team is organized hierarchically and decisions at different levels of the hierarchy are governed by virtual command agents with their own beliefs, goals and plans. The framework supports an interpreter that realizes execution of single or multiagent plans coherently. The framework is implemented and a case study demonstrating the capabilities of the framework is carried out.

Managing a sustainable built environment with a large number of buildings rests on the ability to assess and improve the performance of the building stock over time. Building stock models are cornerstones to the assessment of the combined impact of energy-related building interventions across different spatial and temporal scales. However, such models, particularly those accounting for both physical formulation and social behaviors of the underlying buildings, are still in their infancy. This research strives to more thoroughly examine how buildings perform aggregately in energy usage by focusing on how to tackled three major technical challenges: (1) quantifying building energy performance in an objective and scalable manner, (2) mapping building stock model space to real-world data space, and (3) quantifying and evaluating energy intervention behaviors of a building stock. This thesis hypothesizes that a new paradigm of aggregation of large-scale building stocks can lead to (1) an accurate and efficient intervention analysis model and (2) a functionally comprehensive decision support tool for building stock energy intervention analysis. Specifically, this thesis presents three methodologies. To address the first challenge, this thesis develops a normative building physical energy model that can rapidly estimate single building energy performance with respect to its design and operational characteristics. To address the second challenge, the thesis proposes a statistical procedure using regression and Markov chain Monte Carlo (MCMC) sampling techniques that inverse-estimate building parameters based on building stock energy consumption survey data. The outcomes of this statistical procedure validate the approach of using prototypical buildings for two types of intervention analysis: energy retrofit and demand response. These two cases are implemented in an agent-based modeling and simulation (ABMS) framework to tackle the third challenge. This thesis research contributes to the body of knowledge pertaining to building energy modeling beyond the single building scale. The proposed framework can be used by energy policy makers and utilities for the evaluation of energy retrofit incentives and demand-response program economics.