报告题目:A Good Asymptotic Framework is Important for Constructing Valid Nonparametric Confidence Intervals
主讲人:郭绍俊
报告摘要:In this article we revisit the problem of how to constructbetter nonparametric confidence inter-vals for the conditionalquantile function from an optimization perspective. We applythe fully data-driven bias correction procedure based on localpolynomial smoothing to estimate the conditional quantile. Toaccount for the effect of the estimated bias, we apply an asymptotic framework that the ratio of the bandwidth to the pilotbandwidth tends to some positive constant rather than zero asthe sample size grows. We derive an alternative asymptoticnormality of the proposed bias-corrected quantile estimator aswell as a new asymptotic variance formula. Based on theoretical results, two new pointwise confidence intervals are constructed through resampling strategies. Extensive simulationstudies show that our proposed confidence intervals enjoybetter performance than other competitors in terms of coverageprobabilities and interval lengths and are not sensitive to thechoice of bandwidth. Finally, our proposed procedure is furtherillustrated through UnitedStates' natality birth data in 2017.
郭绍俊简介:现为中国人民大学统计与大数据研究院副教授.2003年本科毕业于山东师范大学,2008年获得中国科学院数学与系统科学研究院理学博士学位.博士毕业后留中国科学院数学与系统科学研究院工作,助理研究员,任期至2016年.2009年- 2010年赴美国普林斯顿大学运筹与金融工程系博士后研究,做高维数据分析方面的研究工作,并于2014-2016年在英国伦敦经济学院统计系做博士后研究,做大维时间序列建模方面的研究.目前主要研究方向有:统计学习;非参数及半参数统计建模;生存分析及函数型数据分析等.
报告时间: 2020年11月19日下午3:00
报告地点:南湖校区教学科研楼507室
主办单位:科研处/数学与统计学院
