报告主题：High-dimensional Robust Factor Analysis (HDRFA)
时间：2023年4月7日13:30-14:30
地点：上川路校区一教218
报告人：何勇
报告内容简介：Factor models have been widely applied in economics and finance, and the well-known heavy-tailedness of macroeconomic/financial data is rarely taken into account in the literature. The PCA method for Factor model performs poorly for heavy-tailed data, and thus all statistical methods relying on PCA for factor model would fail to work well in such case, including but not limited to change points detection. In this talk, I will introduce the existing robust factor analysis methods, namely, the Huber Principal Component Analysis (HPCA), the Quantile Factor Analysis (QFA) and the Robust Two Step (RTS). In recent years, matrix-valued or even high order tensor time series have been common in areas of economics and finance. I will also briefly introduce the existing robust factor analysis tools for well-structured matrix/tensor data, extending the HPCA, QFA and RTS in a proper/clever way. The talk is based on some recent work by our group and we also develop an R package “HDRFA” which is available at CRAN https://cran.r-project.org/web/packages/HDRFA/index.html , and related papers are available at my personal website https://heyongstat.github.io/.
主讲人简介：
何勇，山东大学金融研究院，研究员，山东大学未来青年学者；山东大学学士（2012），复旦大学博士（2017），师从张新生教授；从事金融计量统计、高维统计推断以及机器学习等方面的研究，在国际计量及统计学权威期刊Journal of Econometrics, Journal of Business and Economic Statistics, Biometrics (封面文章), Biostatistics、中国科学：数学等发表研究论文30余篇；主持国家自然科学基金面上项目、青年基金，全国统计科学研究重点项目等，获第一届统计科学技术进步奖（第二位）。担任美国数学评论评论员，及JRSSB, JRSSC, Biometrics, EJS,JMVA等国际知名学术期刊匿名审稿人。