光华讲坛——社会名流与企业家论坛第5450期
主题:Pairwise-rank-likelihood methods for the semiparametric transformation model
主讲人:新加坡国立大学 余涛副教授
主持人:统计学院统计研究中心 林华珍教授
时间:2019年6月20日(星期四)下午2:00-3:00
地点:西南财经大学柳林校区弘远楼408会议室
主办单位:统计研究中心 统计学院 科研处
主讲人简介:
Dr. YU, Tao received his B.S. degree and M.S. in Mathematics and Probability & Statistics from Nankai University in 2001 and 2004 respectively. He obtained his Ph.D. degree from University of Wisconsin-Madison in 2009. He was assistant professor from September 2009 to December 2016 in Department of Statistics and Applied Probability (DSAP) at National University of Singapore (NUS), and now he is associate professor in DSAP at NUS.
余博士于2001年和2004年分别在南开大学获得数学和概率统计学士学位和硕士学位。2009年,他在威斯康辛大学麦迪逊分校获得博士学位。2009年9月至2016年12月,新加坡国立大学统计与应用概率系助理教授,现任新加坡国立大学统计与应用概率系副教授。
主要内容:
In this paper, we study the linear transformation model in the most general setup. This model includes many important and popular models in statistics and econometrics as special cases. Although it has been studied for many years, the methods in the literature either are based on kernel-smoothing techniques or make use of only the ranks of the responses in the estimation of the parametric components. The former approach needs a tuning parameter, which is not easily optimally specified in practice; and the latter approach may be less accurate and computationally expensive. In this paper, we propose a pairwise rank likelihood method and extend it to a score-function-based method. Our methods estimate all the unknown parameters in the linear transformation model, and we explore the theoretical properties of our proposed estimators. Via extensive numerical studies, we demonstrate that our methods are appealing in that the estimators are not only robust to the distribution of the random errors but also in many cases more accurate than those of the existing methods.
本文研究了最常用的线性变换模型。该模型包括许多重要和流行的统计和计量经济学模型作为特殊情况。虽然已经研究了很多年,但是文献中的方法要么基于核平滑技术,要么仅在参数分量的估计中使用响应的等级。前一种方法需要一个调整参数,在实践中不容易最佳地指定;而后一种方法可能不太准确且计算成本高。在本文中,我们提出了一种成对秩可能性方法,并将其扩展到基于分数函数的方法。我们的方法估计了线性变换模型中的所有未知参数,并且我们探索了我们提出的估计量的理论性质。通过广泛的数值研究,我们证明了我们的方法具有吸引力,因为估计量不仅对随机误差的分布具有稳健性,而且在许多情况下比现有方法更准确。