文字
缩放
增大字体
减小字体
香港浸会大学 彭衡副教授等:Local Average Estimation and Inferences for Varying Coecient Models

([西财资讯] 发布于 :2018-05-26 )

光华讲坛——社会名流与企业家论坛第4969

 

主题Local Average Estimation and Inferences for Varying Coecient Models

主讲人:香港浸会大学 彭衡副教授

主持人统计学院 林华珍教授

时间2018年5月28日(星期一10:00-12:00

地点:利记娱乐网柳林校区弘远楼408学术会议室

主办单位:统计研究中心 统计学院 科研处

 

主题Local Average Estimation and Inferences for Varying Coecient Models

主讲人香港浸大学 彭衡副教授

主讲人概况:

彭衡现为香港浸会大学副教授2003-2006年普林斯顿大学运营研究与金融工程学系做助理研究员2006-2014香港浸会大学数学学系做助理教授。其研究兴趣为金融计量经济学、生物信息学、数据分析建模、模型选择、非参数方法等公开发表论文10余篇。

摘要

The varying coecient model is very popular in the application of finance, economics, medical science and many other areas, but the estimation and inference process are quite compute-intensive. This paper presents a local average method to reduce the computation burden. The estimation for the varying coecients is discussed and is extended to the partially linear varying coecient model. Furthermore, three tests are brought out to check whether certain coecient is constant or even signicant. The proposed tests are very easy to implement and their asymptotically distributions under null hypothesis have been deduced. Simulations and real data application are also studied to illustrate the local average method.

 

主题二CoMM: a collaborative mixed model to dissecting genetic contributions to complex traits by leveraging regulatory information

主讲人:香港科技大学 杨灿博士

主讲人概况:

杨灿,现为香港科技大学数学系助理教授,其研究兴趣是统计机器学习、大数据、统计遗传学和基因组学、生物信息学等杨博士的研究重点是对大规模数据分析的新统计和计算方法的发展,包括线性混合模型(LMM)、低阶近似和超高维回归的相互作用检测。他共公开发表文章65篇他的研究论文多发表在高影响力期刊上,如统计年鉴、生物信息学、IEEE模式分析和机器智能、公共科学图书馆遗传学、美国国家科学院学报以及美国人类遗传学杂志。在他对高维数据分析的贡献的基础上,杨博士获得了2012年香港青年科学家奖。

摘要

Genome-wide association studies (GWASs) have been successful in identifying genetic variants associated with complex traits. However, the mechanistic links underlying how these genetic variants affect complex traits remain elusive. A scientific hypothesis is that genetic variants influence complex traits at the organismal level via affecting cellular traits, such as regulating gene expression and altering protein abundance. Although earlier works have already presented some scientific insights about this hypothesis and their findings are very promising, statistical methods that effectively harness multilayer data (e.g., genetic variants, cellular traits and organismal traits) on a large scale for functional and mechanistic exploration are highly demanding. In this article, we propose a collaborative mixed model (CoMM) to dissect genetic contributions to complex traits by leveraging regulatory information in transcriptome data so that the mechanistic role of associated variants could be fully revealed. To demonstrate the advantages of CoMM over existing methods, we conducted extensive simulation studies and also applied CoMM to analyze 25 traits in NFBC1966 and Genetic Epidemiology Research on Aging (GERA) studies. The results indicate that by leveraging regulatory information, CoMM can effectively improve the power of prioritizing risk variants. This is the join work of Xiang Wan, Xinyi Lin, Mengjie Chen, Xiang Zhou and Jin Liu.


☆该资讯已被浏览: 次★

打印本文】 【关闭窗口



XML 地图 | Sitemap 地图