报告主题：High-Dimensional Subgroup Identification under Latent Factor Structures

时间：2024年5月17日13:00-13:45

地点：上川路校区二教105/腾讯会议 会议号：517 4242 8784 密码：6666

报告人：张明娟

报告内容简介：

In the subgroup analysis, the high dependence across features brings enormous challenges in the group identification and consistency of oracle estimations. Motivated by Factor Augmented (sparse) Linear Model (FARM) to bridge dimension reduction and sparse regression together, we propose the Center Augmented FARM (CAFARM) for grouping and sparsity pursuit simultaneously. The CAFARM is flexible to the cross-sectional dependence and inherits the computational complexity with O(nK), contrasting with the O(n2) computational complexity of pairwise penalty, where n is the sample size and K is the number of subgroups. We study the asymptotic properties of its oracle estimators with requirements for the minimal distance between group centroids and put forward corresponding two-step algorithms with convergence of the algorithm. We exhibit the superior accuracy and efficiency of the proposed method through extensive numerical experiments and a real macroeconomic data example. An R packages CAFARM is also provided for implementation.

主讲人简介：

张明娟，上海立信会计金融学院副教授。研究方向高维数据统计推断，主持国家自然科学基金青年项目。