报告题目：Estimation and Inference for Multi-Kink Quantile Regression
报 告 人：钟威教授 厦门大学
报告时间：2020年7月7日 9:00-10:00
报告地点：腾讯会议 ID：850 929 796
点击链接入会，或添加至会议列表：
https://meeting.tencent.com/s/FAnC7BsqUkGF
校内联系人：朱复康 fzhu@jlu.edu.cn
报告摘要：
This article is concerned with parameter estimation, change points detection and statistical inference for a Multi-Kink Quantile Regression (MKQR) model. It assumes different regression forms in different regions of the domain of the threshold covariate but is still continuous at multiple kink thresholds. We propose an iterative segmented quantile regression algorithm for estimating both the regression coefficients and the locations of kink points. It is much more computationally efficient than the grid search algorithm and not sensitive to the initial values. We theoretically demonstrate that the selection consistency of the number of kink points and the asymptotic normality of both regression coefficients and kink locations parameters. The MKQR model is robust to outliers and heavy-tailed errors in the response and more flexible for modelling data with heterogeneous conditional distributions especially when upper or lower quantiles of the response are of interest. Monte Carlo simulations and two real data applications illustrate the excellent performances of the proposed method.
报告人简介：
钟威，现任厦门大学王亚南经济研究院和经济学院统计系教授、博士生导师，经济学院经理助理，国家自然科学基金优秀青年基金获得者（2019），福建省自然科学杰出青年基金获得者（2019）。2012年获得美国宾夕法尼亚州立大学统计学博士学位，2014年和2017年分别破格晋升副教授和教授。主要从事高维数据统计分析和理论、统计学习和数据挖掘算法、计量经济学、统计学和数据科学的应用等领域的研究。在The Annals of Statistics, Journal of the American Statistical Association, Biometrika, Journal of Business & Economic Statistics, Annals of Applied Statistics, Statistica Sinica，中国科学数学等国内外统计学权威期刊发表20多篇论文。担任美国统计协会(ASA)期刊《Statistical Analysis and Data Mining》的副主编。
