講座嘉賓：劉妍巖 教授

講座日期：2019-12-24

講座時間：15:00

報告地點：長安校區(qū) 數(shù)學與信息科學學院學術交流廳

主辦單位：數(shù)學與信息科學學院

講座人簡介：

劉妍巖，武漢大學數(shù)學與統(tǒng)計學院教授，博士生導師。2001年獲武漢大學理學博士學位。主要研究方向為生存分析、半?yún)?shù)統(tǒng)計推斷、高維數(shù)據(jù)統(tǒng)計分析等。主持完成國家自然科學基金以及教育部基金項目6項，正在主持國家自然科學基金面上項目一項，參加完成的成果“風險模型中的統(tǒng)計方法及相關理論與應用” 獲2013年湖北省自然科學獎三等獎（排名第一）。在Journal of Machine Learning Research, Biometrics, Biostatistics, Genetics，Lifetime Data Analysis等期刊發(fā)表SCI研究論文五十余篇。目前擔任中國現(xiàn)場統(tǒng)計學會第十屆理事會常務理事、中國教育統(tǒng)計協(xié)會常務理事、中國工業(yè)統(tǒng)計學會常務理事、中國數(shù)學會女數(shù)學家及西部數(shù)學發(fā)展工作委員會委員。

講座簡介：

Regularization methods for the Cox proportional hazards regression with high-dimensional survival data have been studied extensively in the literature. However, if the models are misspecified, this would result in misleading statistical inference and prediction. To enhance the prediction accuracy for the relative risk and the survival probability of clinical interest, we propose three model averaging approaches for the high-dimensional Cox proportional  hazards regression. Based on the martingale residual process, we define the delete-one crossvalidation process. Further, we propose three novel cross-validation functionals, including the end-time cross-validation, integrated cross-validation, and supremum cross-validation, to achieve more accurate prediction for the risk quantities. The optimal weights for candidate models, without the constraint of summing up to one, can be obtained by minimizing these functionals, respectively. The proposed model averaging approaches can attain the lowest possible prediction loss asymptotically. Furthermore, we develop a greedy model averaging algorithm to overcome the computational obstacle when the dimension is high. The performance of the proposed model averaging procedures is evaluated via extensive simulation studies, showing that our methods have superior prediction accuracy over the existing regularization methods. As an illustration, we apply the proposed methods to the mantle cell lymphoma study.