Frequentist Model Averaging for Undirected and Directed Acyclic Gaussian Graphical Models
报告人: 张新雨(中科院数学与系统科学研究院)
时间:2023-12-07 15:10-17:00
地点:Room 217, Guanghua Building 2
Abstract:
Advances in information technologies have made network data increasingly frequent in economics, finance and genetics, which is often explored by probabilistic graphs models. To estimate the network, we propose optimal model averaging methods. With a set of candidate models varying by graph structures, we propose weights to average the estimates from candidate models. We prove the asymptotic optimality, weight consistency, and parameter consistency of the proposed method. Furthermore, the results from simulations and real data analysis on yeast genetic data and bank’s international liability data show that the proposed methods are promising.
About the Speaker:
张新雨,中科院数学与系统科学研究院研究员。主要从事统计学和计量经济学的理论和应用研究工作,具体研究方向包括模型平均、管理统计、机器学习和经济预测等,担任SCI期刊《Journal of Systems Science & Complexity (JSSC)》领域主编和其他5个国内外重要期刊的编委,是管理科学与工程学会常务理事。
Your participation is warmly welcomed!
欢迎扫码关注北大统计科学中心公众号,了解更多讲座信息!