서브메뉴
검색
Strategies for Communicating Uncertainty in Predictive Systems: Designing Interfaces and Visualizations for Enhanced Data-Driven Decision-Making
Strategies for Communicating Uncertainty in Predictive Systems: Designing Interfaces and Visualizations for Enhanced Data-Driven Decision-Making
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20250211151410
- ISBN
- 9798382762777
- DDC
- 004
- 저자명
- Zhang, Dongping.
- 서명/저자
- Strategies for Communicating Uncertainty in Predictive Systems: Designing Interfaces and Visualizations for Enhanced Data-Driven Decision-Making
- 발행사항
- [Sl] : Northwestern University, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 243 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 85-11, Section: B.
- 주기사항
- Advisor: Hullman, Jessica.
- 학위논문주기
- Thesis (Ph.D.)--Northwestern University, 2024.
- 초록/해제
- 요약The increasing availability of online data and advances in computing technologies have accelerated the development and democratized the application of complex statistical models to assist decision-making. These powerful tools enable researchers and analysts to gain valuable statistical insights. However, the model output often presents interpretation challenges, making it difficult to quantify and visualize uncertainty for informed data-driven decision-making, especially in high-stakes scenarios. To address these challenges, my dissertation presents tools and methods designed to communicate the uncertainty inherent in statistical inferences and predictions commonly seen in real-world applications.I investigate the utility of these techniques in three distinct decision-making contexts. First, I explore the application of Network Hypothetical Outcome Plots (NetHOPs) to support graph-based decision-making, where decision-makers need to reason about uncertainty embedded in complex probabilistic relationships to identify patterns, structural configurations, and cluster occurrences. Second, I examine the utility of Conformal Prediction, a distribution-free uncertainty quantification framework, in supporting AI-advised decision-making, where decision-makers incorporate the uncertainty of AI insights into their own judgment before making decisions. Third, I demonstrate the importance of interface design in supporting strategic decision-making, where decision-makers need to incorporate their anticipation of competitors' actions based on shared information stimuli.My findings demonstrate that effective data-driven decision-making is not only influenced by a model's overarching accuracy but also hinges on decision-makers' capacity to interpret the model output, which is shaped by the way uncertainty is communicated and visualized in predictive interfaces. By addressing the challenges of uncertainty quantification, visualization, and human factors in data-driven decision-making, this dissertation establishes a foundation for future research at the intersection of information visualization, graph analysis, machine learning, and empirical game theory. The insights and techniques contribute to the development of more effective, reliable, and trustworthy predictive systems that can support informed decision-making across a wide range of domains.
- 일반주제명
- Computer science
- 일반주제명
- Communication
- 일반주제명
- Behavioral sciences
- 일반주제명
- Behavioral psychology
- 키워드
- Decision-making
- 키워드
- Deep learning
- 키워드
- Game theory
- 키워드
- Graph theory
- 기타저자
- Northwestern University Technology and Social Behavior
- 기본자료저록
- Dissertations Abstracts International. 85-11B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008250123s2024 us c eng d■001000017161543
■00520250211151410
■006m o d
■007cr#unu||||||||
■020 ▼a9798382762777
■035 ▼a(MiAaPQ)AAI31292670
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a004
■1001 ▼aZhang, Dongping.▼0(orcid)0000-0001-9825-1411
■24510▼aStrategies for Communicating Uncertainty in Predictive Systems: Designing Interfaces and Visualizations for Enhanced Data-Driven Decision-Making
■260 ▼a[Sl]▼bNorthwestern University▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a243 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 85-11, Section: B.
■500 ▼aAdvisor: Hullman, Jessica.
■5021 ▼aThesis (Ph.D.)--Northwestern University, 2024.
■520 ▼aThe increasing availability of online data and advances in computing technologies have accelerated the development and democratized the application of complex statistical models to assist decision-making. These powerful tools enable researchers and analysts to gain valuable statistical insights. However, the model output often presents interpretation challenges, making it difficult to quantify and visualize uncertainty for informed data-driven decision-making, especially in high-stakes scenarios. To address these challenges, my dissertation presents tools and methods designed to communicate the uncertainty inherent in statistical inferences and predictions commonly seen in real-world applications.I investigate the utility of these techniques in three distinct decision-making contexts. First, I explore the application of Network Hypothetical Outcome Plots (NetHOPs) to support graph-based decision-making, where decision-makers need to reason about uncertainty embedded in complex probabilistic relationships to identify patterns, structural configurations, and cluster occurrences. Second, I examine the utility of Conformal Prediction, a distribution-free uncertainty quantification framework, in supporting AI-advised decision-making, where decision-makers incorporate the uncertainty of AI insights into their own judgment before making decisions. Third, I demonstrate the importance of interface design in supporting strategic decision-making, where decision-makers need to incorporate their anticipation of competitors' actions based on shared information stimuli.My findings demonstrate that effective data-driven decision-making is not only influenced by a model's overarching accuracy but also hinges on decision-makers' capacity to interpret the model output, which is shaped by the way uncertainty is communicated and visualized in predictive interfaces. By addressing the challenges of uncertainty quantification, visualization, and human factors in data-driven decision-making, this dissertation establishes a foundation for future research at the intersection of information visualization, graph analysis, machine learning, and empirical game theory. The insights and techniques contribute to the development of more effective, reliable, and trustworthy predictive systems that can support informed decision-making across a wide range of domains.
■590 ▼aSchool code: 0163.
■650 4▼aComputer science
■650 4▼aCommunication
■650 4▼aBehavioral sciences
■650 4▼aBehavioral psychology
■653 ▼aDecision-making
■653 ▼aDeep learning
■653 ▼aGame theory
■653 ▼aGraph theory
■653 ▼aInformation visualization
■653 ▼aUncertainty communication
■690 ▼a0984
■690 ▼a0459
■690 ▼a0602
■690 ▼a0800
■690 ▼a0384
■71020▼aNorthwestern University▼bTechnology and Social Behavior.
■7730 ▼tDissertations Abstracts International▼g85-11B.
■790 ▼a0163
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17161543▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
