서브메뉴
검색
Assumption-Lean Approaches to Modern Statistical Inference
Assumption-Lean Approaches to Modern Statistical Inference
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202103617
- ISBN
- 9798286453030
- DDC
- 310
- 저자명
- Hore, Rohan.
- 서명/저자
- Assumption-Lean Approaches to Modern Statistical Inference
- 발행사항
- [Sl] : The University of Chicago, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 286 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-01, Section: B.
- 주기사항
- Advisor: Foygel Barber, Rina.
- 학위논문주기
- Thesis (Ph.D.)--The University of Chicago, 2025.
- 초록/해제
- 요약In the modern era of machine learning, where increasingly complex models drive decision-making, statistical inference becomes challenging. Assumption-lean inference minimizes reliance on distributional assumptions, ensuring robustness across diverse settings. This thesis explores two key aspects-predictive inference via conformal prediction and conditional independence testing-each providing practical, interpretable solutions to real-world challenges. The first part investigates distribution-free predictive inference. Conformal prediction (CP) constructs prediction intervals with distribution-free guarantees, but its validity relies on exchangeability and only ensures marginal coverage, leading to failures in non-i.i.d. settings, undercoverage for certain subpopulations, or discrepancies when test and training distributions differ. We address these limitations in two steps. First, in survival analysis, where right-censoring restricts observation of true survival times, we propose a novel lower prediction bound using a data-adaptive filter that excludes low censoring times, ensuring valid marginal coverage under mild conditions. Next, we tackle local coverage, aiming for prediction intervals that retain validity for each feature configuration. While exact local coverage is theoretically impossible, we introduce randomly-localized conformal prediction (RLCP), a framework that provides interpretable guarantees for relaxed notions of local validity, such as approximate coverage under smooth covariate shifts or within sufficiently large subpopulations. The second part develops assumption-lean tests for conditional independence. Given the inherent difficulties of testing conditional independence in a distribution-free manner, we propose a test for the conditional independence of X and Y given Z, under the assumption that X is stochastically increasing in Z. Our procedure, PairSwap-ICI, offers significant flexibility while ensuring finite-sample Type I error control and achieving high power against a broad range of alternatives that deviate sufficiently from the null.
- 일반주제명
- Statistics
- 일반주제명
- Applied mathematics
- 기타저자
- The University of Chicago Statistics
- 기본자료저록
- Dissertations Abstracts International. 87-01B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017357917
■00520260202103617
■006m o d
■007cr#unu||||||||
■020 ▼a9798286453030
■035 ▼a(MiAaPQ)AAI32044859
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a310
■1001 ▼aHore, Rohan.▼0(orcid)0000-0001-7670-9700
■24510▼aAssumption-Lean Approaches to Modern Statistical Inference
■260 ▼a[Sl]▼bThe University of Chicago▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a286 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-01, Section: B.
■500 ▼aAdvisor: Foygel Barber, Rina.
■5021 ▼aThesis (Ph.D.)--The University of Chicago, 2025.
■520 ▼aIn the modern era of machine learning, where increasingly complex models drive decision-making, statistical inference becomes challenging. Assumption-lean inference minimizes reliance on distributional assumptions, ensuring robustness across diverse settings. This thesis explores two key aspects-predictive inference via conformal prediction and conditional independence testing-each providing practical, interpretable solutions to real-world challenges. The first part investigates distribution-free predictive inference. Conformal prediction (CP) constructs prediction intervals with distribution-free guarantees, but its validity relies on exchangeability and only ensures marginal coverage, leading to failures in non-i.i.d. settings, undercoverage for certain subpopulations, or discrepancies when test and training distributions differ. We address these limitations in two steps. First, in survival analysis, where right-censoring restricts observation of true survival times, we propose a novel lower prediction bound using a data-adaptive filter that excludes low censoring times, ensuring valid marginal coverage under mild conditions. Next, we tackle local coverage, aiming for prediction intervals that retain validity for each feature configuration. While exact local coverage is theoretically impossible, we introduce randomly-localized conformal prediction (RLCP), a framework that provides interpretable guarantees for relaxed notions of local validity, such as approximate coverage under smooth covariate shifts or within sufficiently large subpopulations. The second part develops assumption-lean tests for conditional independence. Given the inherent difficulties of testing conditional independence in a distribution-free manner, we propose a test for the conditional independence of X and Y given Z, under the assumption that X is stochastically increasing in Z. Our procedure, PairSwap-ICI, offers significant flexibility while ensuring finite-sample Type I error control and achieving high power against a broad range of alternatives that deviate sufficiently from the null.
■590 ▼aSchool code: 0330.
■650 4▼aStatistics
■650 4▼aApplied mathematics
■653 ▼aAssumption-lean inference
■653 ▼aConformal prediction
■653 ▼aDistribution-free inference
■653 ▼aShape-constrained testing
■690 ▼a0463
■690 ▼a0800
■690 ▼a0364
■71020▼aThe University of Chicago▼bStatistics.
■7730 ▼tDissertations Abstracts International▼g87-01B.
■790 ▼a0330
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17357917▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


