主题：Perturbation-Based Tests with Application to the Clayton Model等
主办单位：统计研究中心 统计学院 科研处
主题一：Perturbation-Based Tests with Application to the Clayton Model
现为Canadian Journal of Statistics ，Journal of Statistical Distributions and Applications ， Journal of Biometrics & Biostatistics 和Austin Statistics的副主编。
Perturbation resampling method can be employed to estimate the covariance matrix of an estimator when the estimator is obtained through minimizing a U-process. This perturbation resampling is proposed to establish general tests for the detection of model misspecification or for model checking. The proposed tests enjoy simplicity and a theoretical justification. We apply the proposed method to modify the tests proposed by Shih (1998) for the assessment of Clayton models in multivariate survival analysis, where the asymptotic variance is intractable. The proposed tests present promising performance in the simulation studies and have simpler procedures than the nonparametric bootstrap which can also be applied to approximate the covariance matrix. A colon cancer study further illustrates the proposed methods.
主题二：Causal Inference with Measurement Error in Outcomes
Grace Y. Yi is Professor of Statistics and University Research Chair at the University of Waterloo. Grace received her Ph.D. in Statistics from the University of Toronto in 2000. She is a Fellow of the American Statistical Association and an Elected Member of the International Statistical Institute. She is the Editor-in-Chief of The Canadian Journal of Statistics (2016-2018). She was President of the Biostatistics Section of The Statistical Society of Canada in 2016, and the Founder and Chair of the ﬁrst chapter (Canada Chapter) of The International Chinese Statistical Association. Grace’s research interests focus on developing statistical methodology to address various challenges arising from medical studies, clinical trials, epidemiology ,etc. Recently, her research monograph, “Yi, G. Y. (2017). Statistical Analysis with Measurement Error or Misclassiﬁcation: Strategy, Method and Application”, has been published by Springer. Grace has published a broad range of academic papers in reputable journals, including Journal of The American Statistical Society, Biometrika, Journal of the Royal Statistical Society (Series B), Biometrics, Statistica Sinica, ，etc.Grace was the 2010 winner of the CRM-SSC Prize, an honor awarded in recognition of a statistical scientist’s professional accomplishments in research during the ﬁrst 15 years after having received a doctorate. She was a recipient of the prestigious University Faculty Award granted by the Natural Sciences and Engineering Research Council of Canada (NSERC). Her work with Xianming Tan and Runze Li won The Canadian Journal of Statistics Award in 2016. Grace has supervised many trainees at diﬀerent levels, including post-doctoral fellows and undergraduate internship students. Two of her Ph.D. students (2015 and 2009) received The Pierre Robillard Award, a prestigious award in recognition of the best PhD thesis in the areas of probability and statistics defended in Canada each year (the award was established in 1978). Grace has served as an Associate Editor for statistical journals, including The Canadian Journal of Statistics, Journal of the Royal Statistical Society (Series C), The Journal of Applied Probability, Statistics in Biosciencs, STAT, and Biostatistics and Epidemiology. She has served on various professional committees and organizations, including the Mathematics and Statistics Evaluation Group of Discovery Grants Program at NSERC, the Fisher Lecture Committee of the American Statistical Association (ASA), the Development Committee for Canadian Statistical Science Institute (CANSSI), the Advisory Board of the Eastern North American Region (ENAR), and the Committee of the Distinguished Lecture in Statistical Sciences at the Fields Institute.
Inverse probability weighting (IPW) estimation has been popularly used to consistently estimate the average treatment effect (ATE). Its validity, however, is challenged by the presence of error-prone variables. In application, measurement error is ubiquitously present in data collection due to various reasons. Naively ignoring measurement error effects usually yields biased inference results. In this talk, I will discuss the IPW estimation with mismeasured outcome variables. The impact of measurement error for both continuous and discrete outcome variables will be examined. I will describe estimation procedures with the outcome misclassification effects accommodated. Consistency and efficiency will be investigated.
Numerical studies will be reported to assess the performance of the proposed methods.
主题三：Disentangling and Assessing Uncertainties in Multiperiod Corporate Default Risk Prediction
汤琤咏，现为天普大学统计科学系副教授，2008年在爱荷华州立大学取得统计学博士学位，在他博士期间，他的Major Professor是 Dr. Song Xi Chen。他也是天普大学统计学系Graduate Programs的主任。他的研究兴趣包括：统计方法，高维数据分析，经验似然，纵向数据分析，金融统计与计量经济学，缺失数据的抽样统计与分析，非参数和半参数统计方法等。已公开发表期刊论文29篇。主持科研项目6项。他是International Statistical Institute (ISI)的Elected Member；The Royal Statistical Society的Fellow；他是ASA ，IMS和ICSA的Member。
Measuring credit risks for individual companies, industrial segments, and market systems is fundamentally and broadly important in economics, ﬁnance and beyond. For such a purpose, various quantitative methods have been developed to predictively assess the probabilities of companies going default in future. However, as a more diﬃcult yet crucial problem, evaluating the uncertainties associated with the default predictions remains little explored. In this paper, we develop, for the ﬁrst time in the scenario of default predictions, a procedure for quantifying the level of associated uncertainties by carefully disentangling multiple contributing sources. Our framework eﬀectively incorporates broad information from historical default data, ﬁnancial records, and economic environmental conditions by a) characterizing the default mechanism, and b) capturing the future dynamics of various features contributing to the default mechanism. Our development of the framework overcomes major challenges in this tremendously large scale statistical inference problem and makes it practically feasible by using parsimonious models, innovative methods, and modern computational facilities. By appropriately predicting the market-wise total number of defaults and assessing the associated uncertainties, our method can eﬀectively evaluate the aggregated market credit risk level. Upon analyzing a US market data set with our method, we demonstrate that the level of uncertainties associated with default risk assessments is indeed substantial. More importantly and informatively, we also ﬁnd that the level of uncertainties associated with the default risk predictions is correlated with the level of default risks, indicating potential for beneﬁting practical applications including improving the accuracy of default risk assessments. This is a joint work with Miao Yuan, Yili Hong, and Jian Yang.
主题四：Two-sample functional linear models
梁华--乔治•华盛顿大学统计系教授, 曾任美国罗切斯特大学医学院教授。出版英文学术著作2部，发表学术论文 150 多篇，其中24篇在 Annals of Statistics，Biometrika，JASA，和 JRSSB。他主持(了)8项美国国家科学基金会(NSF)以及美国国立卫生研究院(NIH)的研究项目。还主持1项海外-港澳学者研究基金(原杰出青年基金B类, 2013-2019)。
We study two-sample functional linear regression with a scaling transformation of regression functions. We consider estimation of the intercept, the slope function and the scalar parameter based on the functional principal component analysis. We also establish the rates of convergence for the estimator of the slope function, which is shown to be optimal in a minimax sense under certain smoothness assumptions. We further investigate semiparametric efficiency for the estimation of the scalar parameter and hypothesis testing. We also extend the proposed method to sparsely and irregularly sampled functional data and establish the consistency for the estimators of the scalar and the slope function. We evaluate numerical performance of the proposed methods through simulation studies and illustrate their utility via analysis of an AIDS data set.