CVX软件包在统计实验教学中的应用

Application of CVX Software Package in Statistical Experiment Teaching

回归分析是高校统计学的专业必修课,关于模型的变量选择又是该门课程的重点内容。传统的变量选择方法具有很大的局限性。文章基于CVX凸优化包,给出了线性回归模型、分位数回归模型和复合分位数回归模型中变量选择的算法。通过模拟计算说明了该算法的可行性和有效性。

Regression analysis is a compulsory subject of statistics in college and the variable selection of model is the key content of this course. The traditional variable selection method has a lot of limitations. Based on the software package of CVX in Maflab, we propose an optimization algorithm of variable selection in linear regres- sion model, quanti｜e regression model and composite quantile regression model. The simulation study presents the feasibility and validity of the proposed algorithm.