Group penalized multinomial logit models and stock return direction prediction

报 告 人：胡雪梅 教授(重庆工商大学)

时    间：2024年7月10日(周三) 15:00--16:00

地    点：东校区博学楼506

邀 请 人：胡  尧 教授

报告摘要:Multinomial logit model (MLM)  has been proposed as  the most frequently regression model for multi-category response and the widely used functional form for discrete choice probabilities. To deal with correlated data, in this paper we propose G-LASSO/G-SCAD/G-MCP penalized MLM to exert class discovery and class prediction for multi-category classification problems. Firstly, we develop a group coordinate descent algorithm  to simultaneously  complete group  selection and   group estimation,  and   prove its convergence under mild conditions. Secondly, we apply the training set and  group estimations to obtain class probability estimators, choose the Bayes classifier  to identify class index information, and introduce the testing set and a few measures to assess  multi-category prediction performance. Simulations show that the proposed methods outperform LASSO/SCAD/MCP penalized MLM, 3 deep learning methods and 3 machine learning methods in terms of Kappa, PDI, Optimal or Average Accuracy. Finally, we combine group penalized MLM with 58 technical indicators to predict up trends, sideways trends and down trends for stock returns, and show that the proposed methods outperform the other 9 methods  in terms of  Accuracy, PDI, Kappa and HUM. Therefore,  the proposed  method can not  only accommodate the correlation information, but also improve multi-category prediction performance by shrinking group coefficients.

报告人简介：

胡雪梅, 重庆工商大学我院教授, 成渝地区双城经济圈建设研究院(教育部人文社科重点研究基地)博士生导师, 伦敦政治经济学院国家公派访问学者,重庆经开区经济运行局挂职副局长,中科院数学与系统科学研究院控制论国家重点实验室系统科学博士后, “第五批重庆市高等学校优秀人才支持计划”人选, 重庆市“统计学”研究生导师团队负责人,《随机过程》市级一流线下课程负责人. 目前已对半变系数模型的统计推断、半参数模型的经验似然、随机扩散方程的稳健推断、高维数据模型的统计学习等展开了系统研究, 在IEEE Transaction on Information Theory、Journal of Multivariate Analysis、Expert Systems with Applications、Statistical Papers、North American Journal of Economics and Finance、Soft Computing、Journal of Forecasting等学术期刊上发表论文50多篇, 其中SCI/SSCI收录30篇, 主持完成1项国家自然科学基金、1项教育部人文社科项目、4项重庆市科委项目、1项重庆市社科规划项目、3项重庆市教委项目(其中1项重大), 参与获得1项重庆市科学技术奖二等奖, 出版学术专著《高维统计模型的估计理论与模型识别》和《高维数据模型的统计学习方法与预测精度评估》,参编英文专著Investment Strategies-New Advances and Challenges的章节为Aspects Regarding a Deep Understanding of the Prediction for Stock Market Movements.

欢迎广大师生参加！