报告时间:2026年11月19日,周四下午3:00-4:00
地点:36-316
报告人: 许文超 上海对外经贸大学
摘要:We propose a functional Kolmogorov-Smirnov (KS) learning framework for binary scoring with functional covariates. Functional principal component analysis reduces data dimensionality, while an adaptive group Lasso penalty enables simultaneous selection and estimation of multiple functional predictors. Simulation studies demonstrate that the proposed method exhibits superior performance compared with existing methods in terms of the KS statistic, regression function estimation, and variable selection. An application to aircraft QAR data illustrates the method's ability to identify key risk factors and construct interpretable early-warning scores for hard landings.
个人简介:许文超,华东师范大学统计学博士,中国科学院数学与系统科学研究院博士后,现为上海对外经贸大学统计与数据科学学院教授、硕士生导师。主要研究方向为函数型数据分析、模型平均。在Econometric Theory、Statistica Sinica、《中国科学(数学)》等国内外知名杂志上接收或发表论文十余篇,主持并完成了国家自然科学基金青年基金一项。