102 219

Cited 0 times in

Stability selection for LASSO with weights based on AUC

DC Field Value Language
dc.contributor.author권용한-
dc.date.accessioned2022-08-23T01:52:56Z-
dc.date.available2022-08-23T01:52:56Z-
dc.date.issued2022-02-
dc.identifier.urihttps://ir.ymlib.yonsei.ac.kr/handle/22282913/189758-
dc.description.abstractBased on stability selection (Meinshaisen and Buhlmann, 2010), we propose to select variables by weighting them using area under the receiver operating characteristic curves (AUC) from additional modeling rather than simply treating all variables equally. We evaluated the proposed method compared with the conventional methods through an extensive simulation study to measure the true positive rate (TPR), positive predictive value (PPV), and stability of variable selection. We also assessed the method’s prediction ability on a validation set. The performance of the proposed method was comparable to that of the conventional methods in terms of TPR, PPV, and stability. The AUCs on the validation set for the proposed method were always higher. A major advantage of the proposed method is that it enables researchers to select variables intuitively using relatively simple parameter settings. 본 연구는 stability selection (Meinshaisen and Buhlmann, 2010)을 기반으로 하여 단순히 모든 변수를 동등하게 취급하기보다 추가 모델링에서의 AUC를 통한 가중치를 부여하여 변수를 선택할 것을 제안한다. 제안된 방법은 TPR(True Positive Rate), PPV(Positive predictive Value) 및 변수 선택의 안정성을 측정하기 위해 광범위한 시뮬레이션 연구를 통해 기존 방법과 비교 평가했다. 검증 세트에 대한 예측 능력도 평가했다. 제안된 방법의 성능은 TPR, PPV 및 안정성 측면에서 기존 방법과 매우 유사하였다. 제안된 방법에 대한 검증 세트의 AUC는 항상 더 높았다. 제안된 방법의 가장 큰 장점은 연구자가 비교적 간단한 매개변수 설정으로 직관적으로 변수를 선택할 수 있다는 것이다.-
dc.description.statementOfResponsibilityopen-
dc.publisherGraduate School, Yonsei University-
dc.rightsCC BY-NC-ND 2.0 KR-
dc.titleStability selection for LASSO with weights based on AUC-
dc.title.alternativeAUC에 기반한 가중치를 사용한 LASSO에 대한 안정성 선택-
dc.typeThesis-
dc.contributor.collegeCollege of Medicine (의과대학)-
dc.contributor.departmentOthers (기타)-
dc.description.degree석사-
dc.contributor.alternativeNameKwon, Yonghan-
dc.type.localThesis-
Appears in Collections:
1. College of Medicine (의과대학) > Others (기타) > 2. Thesis

qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.