摘要
为了提升在协变量呈现非线性特征时数据的预测精确度,提出了二元贝叶斯可加分位数回归模型.这一方法尤其适用于那些传统线性模型难以应对的复杂情况.为了全面评估该模型的性能,引入了贝叶斯准则(包括DIC与LPML)以及混淆矩阵作为评价指标.通过模拟研究以及对实际数据的深入应用,所得结果充分证实了该模型的有效性.
In order to improve the prediction accuracy of the data when the covariate presents nonlinear characteristics,a binary Bayesian additive quantile regression model is proposed,which is especially suitable for those complex situations that are difficult to cope with by traditional linear models.In order to comprehensively evaluate the performance of the model,the Bayesian criterion(including DIC and LPML)and confusion matrix as evaluation indicators are introduced.Through simulation studies and in-depth application of actual data,the results fully confirm the effectiveness of the proposed model.
作者
姬永刚
朱壮
周茂袁
Ji Yonggang;Zhu Zhuang;Zhou Maoyuan(College of Science,Civil Aviation University of China,Tianjin 300300,China)
出处
《南开大学学报(自然科学版)》
北大核心
2025年第2期107-114,共8页
Journal of Nankai University(Natural Sience)
基金
中国民航大学科研启动基金(2012QD09X)。
关键词
非线性
二元
可加分位数
nonlinear
binary
additive quantile
作者简介
姬永刚(1984-),男,山东淄博人,副教授,研究方向:贝叶斯非参数统计;通讯作者:周茂袁(1980-),男,山东青岛人,教授,研究方向:统计与物理.E-mail:yitangzhang@163.com。
