摘要：AbstractThis paper considers the set-valued system identification in retrospective cohort study and proposes several novel methods based on retrospective likelihood function. We theoretically analyze the difference between retrospective likelihood and traditional prospective likelihood. A general framework is proposed for set-valued system identification in retrospective cohort study and applied to two common problems in Genome-Wide Association Studies(GWAS). Extensive simulations show that new methods perform better than traditional methods in terms of parameter estimation errors.
关键词：KeywordsSet-valued systemretrospective likelihoodsystem identificationGenome-Wide Association Studies
