Dynamic, convex, and robust optimization with Bayesian learning for response-guided dosing

Thesis (Ph.D.)--University of Washington, 2016-08

Precautionary principles: a jurisdiction-free framework for decision-making under risk

Fundamental principles of precaution are legal maxims that ask for preventive actions, perhaps as contingent interim measures while relevant information about causality and harm remains unavailable, to minimize the societal impact of potentially severe or irreversible outcomes. Such principles do not explain how to make choices or how to identify what is protective when incomplete and inconsistent scientific evidence of causation characterizes the potential hazards. Rather, they entrust lower jurisdictions, such as agencies or authorities, to make current decisions while recognizing that future information can contradict the scientific basis that supported the initial decision. After reviewing and synthesizing national and international legal aspects of precautionary principles, this paper addresses the key question: How can society manage potentially severe, irreversible or serious environmental outcomes when variability, uncertainty, and limited causal knowledge characterize their decision-making? A decision-analytic solution is outlined that focuses on risky decisions and accounts for prior states of information and scientific beliefs that can be updated as subsequent information becomes available. As a practical and established approach to causal reasoning and decision-making under risk, inherent to precautionary decision-making, these (Bayesian) methods help decision-makers and stakeholders because they formally account for probabilistic outcomes, new information, and are consistent and replicable. Rational choice of an action from among various alternatives-defined as a choice that makes preferred consequences more likely-requires accounting for costs, benefits and the change in risks associated with each candidate action. Decisions under any form of the precautionary principle reviewed must account for the contingent nature of scientific information, creating a link to the decision-analytic principle of expected value of information (VOI), to show the relevance of new information, relative to the initial ( and smaller) set of data on which the decision was based. We exemplify this seemingly simple situation using risk management of BSE. As an integral aspect of causal analysis under risk, the methods developed in this paper permit the addition of non-linear, hormetic dose-response models to the current set of regulatory defaults such as the linear, non-threshold models. This increase in the number of defaults is an important improvement because most of the variants of the precautionary principle require cost-benefit balancing. Specifically, increasing the set of causal defaults accounts for beneficial effects at very low doses. We also show and conclude that quantitative risk assessment dominates qualitative risk assessment, supporting the extension of the set of default causal models.

Vector Error-Correction Models in a consumer packaged goods category forecasting decision support system

A framework for developing marketing category management decision support systems (DSS) based upon the Bayesian Vector Autoregressive (BVAR) model is extended. Since the BVAR model is vulnerable to permanent and temporary shifts in purchasing patterns over time, a form that can correct for the shifts and still provide the other advantages of the BVAR is a Bayesian Vector Error-Correction Model (BVECM). We present the mechanics of extending the DSS to move from a BVAR model to the BVECM model for the category management problem. Several additional iterative steps are required in the DSS to allow the decision maker to arrive at the best forecast possible. The revised marketing DSS framework and model fitting procedures are described. Validation is conducted on a sample problem.

Collaborative decision-making processes for maintaining biodiversity: Two Australian case studies

There have been many models developed by scientists to assist decision-makers in making socio-economic and environmental decisions. It is now recognised that there is a shift in the dominant paradigm to making decisions with stakeholders, rather than making decisions for stakeholders. Our paper investigates two case studies where group model building has been undertaken for maintaining biodiversity in Australia. The first case study focuses on preservation and management of green spaces and biodiversity in metropolitan Melbourne under the umbrella of the Melbourne 2030 planning strategy. A geographical information system is used to collate a number of spatial datasets encompassing a range of cultural and natural assets data layers including: existing open spaces, waterways, threatened fauna and flora, ecological vegetation covers, registered cultural heritage sites, and existing land parcel zoning. Group model building is incorporated into the study through eliciting weightings and ratings of importance for each datasets from urban planners to formulate different urban green system scenarios. The second case study focuses on modelling ecoregions from spatial datasets for the state of Queensland. The modelling combines collaborative expert knowledge and a vast amount of environmental data to build biogeographical classifications of regions. An information elicitation process is used to capture expert knowledge of ecoregions as geographical descriptions, and to transform this into prior probability distributions that characterise regions in terms of environmental variables. This prior information is combined with measured data on the environmental variables within a Bayesian modelling technique to produce the final classified regions. We describe how linked views between descriptive information, mapping and statistical plots are used to decide upon representative regions that satisfy a number of criteria for biodiversity and conservation. This paper discusses the advantages and problems encountered when undertaking group model building. Future research will extend the group model building approach to include interested individuals and community groups.

Enabling innovation:knowledge control, sharing and coordination boundaries in UK and German manufacturing

We investigate how boundaries in knowledge control, sharing and co-ordination influence UK and German manufacturing firms’ innovation intensity (an indicator of the volume of product change) and product life (an indicator of the pace of generational change). In general UK plants more commonly face knowledge control boundaries related to plant ownership or control, while German plants more commonly face boundaries related to knowledge sharing and knowledge co-ordination between functional groups. Our empirical results emphasise the importance of the strategic management of innovation. Knowledge control boundaries – related to external ownership, group membership and decision making autonomy – have a weak negative influence on plants’ innovation outcomes. Strategic decisions relating to multifunctional working and networking are found to be more important in overcoming knowledge sharing and co-ordination boundaries. Knowledge sharing boundaries, related to plant or company boundaries, prove most important where a plant has no in-house R&D capability. Knowledge co-ordination boundaries related to functional or multi-functional working have strong but differential effects on different innovation output measures: functional boundaries increase product life in both countries, and in Germany maintaining functional boundaries is also associated with increased innovation intensity.

Application of Naïve Bayesian sequential analysis to primary care optometry

The objective of this study was to investigate the effects of circularity, comorbidity, prevalence and presentation variation on the accuracy of differential diagnoses made in optometric primary care using a modified form of naïve Bayesian sequential analysis. No such investigation has ever been reported before. Data were collected for 1422 cases seen over one year. Positive test outcomes were recorded for case history (ethnicity, age, symptoms and ocular and medical history) and clinical signs in relation to each diagnosis. For this reason only positive likelihood ratios were used for this modified form of Bayesian analysis that was carried out with Laplacian correction and Chi-square filtration. Accuracy was expressed as the percentage of cases for which the diagnoses made by the clinician appeared at the top of a list generated by Bayesian analysis. Preliminary analyses were carried out on 10 diagnoses and 15 test outcomes. Accuracy of 100% was achieved in the absence of presentation variation but dropped by 6% when variation existed. Circularity artificially elevated accuracy by 0.5%. Surprisingly, removal of Chi-square filtering increased accuracy by 0.4%. Decision tree analysis showed that accuracy was influenced primarily by prevalence followed by presentation variation and comorbidity. Analysis of 35 diagnoses and 105 test outcomes followed. This explored the use of positive likelihood ratios, derived from the case history, to recommend signs to look for. Accuracy of 72% was achieved when all clinical signs were entered. The drop in accuracy, compared to the preliminary analysis, was attributed to the fact that some diagnoses lacked strong diagnostic signs; the accuracy increased by 1% when only recommended signs were entered. Chi-square filtering improved recommended test selection. Decision tree analysis showed that accuracy again influenced primarily by prevalence, followed by comorbidity and presentation variation. Future work will explore the use of likelihood ratios based on positive and negative test findings prior to considering naïve Bayesian analysis as a form of artificial intelligence in optometric practice.

On the Quality of Decision Functions in Pattern Recognition

The problem of decision functions quality in pattern recognition is considered. An overview of the approaches to the solution of this problem is given. Within the Bayesian framework, we suggest an approach based on the Bayesian interval estimates of quality on a finite set of events.

A world full of surprises:Bayesian theory of surprise to quantify degrees of uncertainty

In the specific area of software engineering (SE) for self-adaptive systems (SASs) there is a growing research awareness about the synergy between SE and artificial intelligence (AI). However, just few significant results have been published so far. In this paper, we propose a novel and formal Bayesian definition of surprise as the basis for quantitative analysis to measure degrees of uncertainty and deviations of self-adaptive systems from normal behavior. A surprise measures how observed data affects the models or assumptions of the world during runtime. The key idea is that a "surprising" event can be defined as one that causes a large divergence between the belief distributions prior to and posterior to the event occurring. In such a case the system may decide either to adapt accordingly or to flag that an abnormal situation is happening. In this paper, we discuss possible applications of Bayesian theory of surprise for the case of self-adaptive systems using Bayesian dynamic decision networks. Copyright © 2014 ACM.

How organizational cognitive neuroscience can deepen understanding of managerial decision-making:a review of the recent literature and future directions

There is growing interest in exploring the potential links between human biology and management and organization studies, which is bringing greater attention to bear on the place of mental processes in explaining human behaviour and effectiveness. The authors define this new field as organizational cognitive neuroscience (OCN), which is in the exploratory phase of its emergence and diffusion. It is clear that there are methodological debates and issues associated with OCN research, and the aim of this paper is to illuminate these concerns, and provide a roadmap for rigorous and relevant future work in the area. To this end, the current reach of OCN is investigated by the systematic review methodology, revealing three clusters of activity, covering the fields of economics, marketing and organizational behaviour. Among these clusters, organizational behaviour seems to be an outlier, owing to its far greater variety of empirical work, which the authors argue is largely a result of the plurality of research methods that have taken root within this field. Nevertheless, all three clusters contribute to a greater understanding of the biological mechanisms that mediate choice and decision-making. The paper concludes that OCN research has already provided important insights regarding the boundaries surrounding human freedom to act in various domains and, in turn, self-determination to influence the workplace. However, there is much to be done, and emerging research of significant interest is highlighted.

Essays in environmental economic valuation and decision making in the presence of an environmental disaster

The first essay developed a respondent model of Bayesian updating for a double-bound dichotomous choice (DB-DC) contingent valuation methodology. I demonstrated by way of data simulations that current DB-DC identifications of true willingness-to-pay (WTP) may often fail given this respondent Bayesian updating context. Further simulations demonstrated that a simple extension of current DB-DC identifications derived explicitly from the Bayesian updating behavioral model can correct for much of the WTP bias. Additional results provided caution to viewing respondents as acting strategically toward the second bid. Finally, an empirical application confirmed the simulation outcomes. The second essay applied a hedonic property value model to a unique water quality (WQ) dataset for a year-round, urban, and coastal housing market in South Florida, and found evidence that various WQ measures affect waterfront housing prices in this setting. However, the results indicated that this relationship is not consistent across any of the six particular WQ variables used, and is furthermore dependent upon the specific descriptive statistic employed to represent the WQ measure in the empirical analysis. These results continue to underscore the need to better understand both the WQ measure and its statistical form homebuyers use in making their purchase decision. The third essay addressed a limitation to existing hurricane evacuation modeling aspects by developing a dynamic model of hurricane evacuation behavior. A household's evacuation decision was framed as an optimal stopping problem where every potential evacuation time period prior to the actual hurricane landfall, the household's optimal choice is to either evacuate, or to wait one more time period for a revised hurricane forecast. A hypothetical two-period model of evacuation and a realistic multi-period model of evacuation that incorporates actual forecast and evacuation cost data for my designated Gulf of Mexico region were developed for the dynamic analysis. Results from the multi-period model were calibrated with existing evacuation timing data from a number of hurricanes. Given the calibrated dynamic framework, a number of policy questions that plausibly affect the timing of household evacuations were analyzed, and a deeper understanding of existing empirical outcomes in regard to the timing of the evacuation decision was achieved.

The boundaries of acceptance: Do expectations for the complainant's psychological injury affect legal decisions in a sexual harassment case

Research indicates that people engaged in legal decision-making use a host of biases and preconceptions to guide their decisions about whether the evidence presented to them is reasonable. However, few theories address how such expectations affect legal decision-makers. The present study attempted to determine if social judgment theory (SJT) can explain how and when legal decision-makers rely on expectations for the complainant's psychological injury in a hostile environment sexual harassment case. Two experiments provided undergraduate participants with a written summary of a hostile work environment allegation that first manipulated participants' expectations about reasonable psychological injuries (mild v. severe), and then presented them with actual severity levels of psychological injury (ranging from minimal to extreme). Experiment 1 (N = 295) hypothesized and found that participants who expected severe injuries perceived a greater range of psychological injuries to be reasonable than participants expecting mild injury. Experiment 2 ( N = 202) used similar methodology and investigated whether perceived reasonableness for the injury allegations affected legal decisions. Experiment 2 hypothesized that participants expecting severe psychological injury should render more pro-complainant decisions than participants expecting mild psychological injury. This result should be most pronounced when participants receive a moderate injury allegation, since this allegation was perceived as reasonable by participants expecting severe injury, but unreasonable by participants expecting mild injury. Consistent with SJT, participants who received a moderate injury but expected a severe injury found more liability than participants who received a moderate injury but expected a mild injury. Inconsistent with SJT, participants' expectations did not affect their compensatory damage decisions. In fact, more severe injury allegations increased damage awards regardless of participants' expectations. Although the results provide mixed support for applying SJT to legal decisions in sexual harassment cases, they emphasize the continuing role of oft-unstudied extra-legal factors (juror's expectations and psychological injury severity) on legal decisions.

Developing brownfields ranking models using decision analytic methods

The redevelopment of Brownfields has taken off in the 1990s, supported by federal and state incentives, and largely accomplished by local initiatives. Brownfields redevelopment has several associated benefits. These include the revitalization of inner-city neighborhoods, creation of jobs, stimulation of tax revenues, greater protection of public health and natural resources, the renewal and reuse existing civil infrastructure and Greenfields protection. While these benefits are numerous, the obstacles to Brownfields redevelopment are also very much alive. Redevelopment issues typically embrace a host of financial and legal liability concerns, technical and economic constraints, competing objectives, and uncertainties arising from inadequate site information. Because the resources for Brownfields redevelopment are usually limited, local programs will require creativity in addressing these existing obstacles in a manner that extends their limited resources for returning Brownfields to productive uses. Such programs may benefit from a structured and defensible decision framework to prioritize sites for redevelopment: one that incorporates the desired objectives, corresponding variables and uncertainties associated with Brownfields redevelopment. This thesis demonstrates the use of a decision analytic tool, Bayesian Influence Diagrams, and related decision analytic tools in developing quantitative decision models to evaluate and rank Brownfields sites on the basis of their redevelopment potential.

Essays in Environmental Economic Valuation and Decision Making in the Presence of an Environmental Disaster

The first essay developed a respondent model of Bayesian updating for a double-bound dichotomous choice (DB-DC) contingent valuation methodology. I demonstrated by way of data simulations that current DB-DC identifications of true willingness-to-pay (WTP) may often fail given this respondent Bayesian updating context. Further simulations demonstrated that a simple extension of current DB-DC identifications derived explicitly from the Bayesian updating behavioral model can correct for much of the WTP bias. Additional results provided caution to viewing respondents as acting strategically toward the second bid. Finally, an empirical application confirmed the simulation outcomes. The second essay applied a hedonic property value model to a unique water quality (WQ) dataset for a year-round, urban, and coastal housing market in South Florida, and found evidence that various WQ measures affect waterfront housing prices in this setting. However, the results indicated that this relationship is not consistent across any of the six particular WQ variables used, and is furthermore dependent upon the specific descriptive statistic employed to represent the WQ measure in the empirical analysis. These results continue to underscore the need to better understand both the WQ measure and its statistical form homebuyers use in making their purchase decision. The third essay addressed a limitation to existing hurricane evacuation modeling aspects by developing a dynamic model of hurricane evacuation behavior. A household’s evacuation decision was framed as an optimal stopping problem where every potential evacuation time period prior to the actual hurricane landfall, the household’s optimal choice is to either evacuate, or to wait one more time period for a revised hurricane forecast. A hypothetical two-period model of evacuation and a realistic multi-period model of evacuation that incorporates actual forecast and evacuation cost data for my designated Gulf of Mexico region were developed for the dynamic analysis. Results from the multi-period model were calibrated with existing evacuation timing data from a number of hurricanes. Given the calibrated dynamic framework, a number of policy questions that plausibly affect the timing of household evacuations were analyzed, and a deeper understanding of existing empirical outcomes in regard to the timing of the evacuation decision was achieved.

Essays in Empirical Operations Management: Bayesian Learning of Service Quality and Structural Estimation of Complementary Product Pricing and Inventory Management

This dissertation contributes to the rapidly growing empirical research area in the field of operations management. It contains two essays, tackling two different sets of operations management questions which are motivated by and built on field data sets from two very different industries --- air cargo logistics and retailing.

The first essay, based on the data set obtained from a world leading third-party logistics company, develops a novel and general Bayesian hierarchical learning framework for estimating customers' spillover learning, that is, customers' learning about the quality of a service (or product) from their previous experiences with similar yet not identical services. We then apply our model to the data set to study how customers' experiences from shipping on a particular route affect their future decisions about shipping not only on that route, but also on other routes serviced by the same logistics company. We find that customers indeed borrow experiences from similar but different services to update their quality beliefs that determine future purchase decisions. Also, service quality beliefs have a significant impact on their future purchasing decisions. Moreover, customers are risk averse; they are averse to not only experience variability but also belief uncertainty (i.e., customer's uncertainty about their beliefs). Finally, belief uncertainty affects customers' utilities more compared to experience variability.

The second essay is based on a data set obtained from a large Chinese supermarket chain, which contains sales as well as both wholesale and retail prices of un-packaged perishable vegetables. Recognizing the special characteristics of this particularly product category, we develop a structural estimation model in a discrete-continuous choice model framework. Building on this framework, we then study an optimization model for joint pricing and inventory management strategies of multiple products, which aims at improving the company's profit from direct sales and at the same time reducing food waste and thus improving social welfare.

Collectively, the studies in this dissertation provide useful modeling ideas, decision tools, insights, and guidance for firms to utilize vast sales and operations data to devise more effective business strategies.

Towards a knowledge-based approach for effective decision-making in railway safety

Purpose – This paper aims to contribute towards understanding how safety knowledge can be elicited from railway experts for the purposes of supporting effective decision-making. Design/methodology/approach – A consortium of safety experts from across the British railway industry is formed. Collaborative modelling of the knowledge domain is used as an approach to the elicitation of safety knowledge from experts. From this, a series of knowledge models is derived to inform decision-making. This is achieved by using Bayesian networks as a knowledge modelling scheme, underpinning a Safety Prognosis tool to serve meaningful prognostics information and visualise such information to predict safety violations. Findings – Collaborative modelling of safety-critical knowledge is a valid approach to knowledge elicitation and its sharing across the railway industry. This approach overcomes some of the key limitations of existing approaches to knowledge elicitation. Such models become an effective tool for prediction of safety cases by using railway data. This is demonstrated using passenger–train interaction safety data. Practical implications – This study contributes to practice in two main directions: by documenting an effective approach to knowledge elicitation and knowledge sharing, while also helping the transport industry to understand safety. Social implications – By supporting the railway industry in their efforts to understand safety, this research has the potential to benefit railway passengers, staff and communities in general, which is a priority for the transport sector. Originality/value – This research applies a knowledge elicitation approach to understanding safety based on collaborative modelling, which is a novel approach in the context of transport.