| 线性模型中变量和变换的同时选择 |
| |
| 引用本文： | 吕纯濂,陈舜华,杨勇杰.线性模型中变量和变换的同时选择[J].数值计算与计算机应用,2005,26(1):26-35. |
| |
| 作者姓名： | 吕纯濂 陈舜华 杨勇杰 |
| |
| 作者单位： | 1. 南京信息工程大学数学系,南京,210044;南京金肯学院,南京,211156
2. 南京信息工程大学应用气象系,南京,210044 |
| |
| 摘 要： | 变量选择和变换选择是线性模型中的两个不同的问题.把这两个过程结合起来同时进行,将是很有意义的.由于近来在计算技术方面的发展,这种同时进行的过程现在是可行的.本文提出了在线性模型中的变数和变换同时进行选择的两个方法.节(?)个方法是(?)个纯粹的同时选择过程.第二个方法适用于具有较多预报因子的数据集,也提出了(?)个对于同时选择的向后删除过程.这两个方法皆以贝叶斯模型选择准则为基础.用(?)个实例来说明这(?)方法.
|
| 关 键 词： | 贝叶斯准则 Box-Cox变换 校准数 比较得分 向后选择 |
| 修稿时间： | 2003年9月5日 |
SIMULTANEOUS SELECTION FOR VARIABLES AND TRANSFORMATION IN LINEAR MODEL |
| |
| Lü Chunlian,Chen Shunhua,Yang Yongjie.SIMULTANEOUS SELECTION FOR VARIABLES AND TRANSFORMATION IN LINEAR MODEL[J].Journal on Numerical Methods and Computer Applications,2005,26(1):26-35. |
| |
| Authors: | Lü Chunlian Chen Shunhua Yang Yongjie |
| |
| Abstract: | Variable selection and transformation selection are two different problems in the linear model. It is very interesting to combine these two procedures in simultaneous one. Due to recent developments in computing technology, such a simultaneous procedure is now feasible. Two procedures for variable and transformation simultaneous selection are proposed in the linear model. The first procedure is a pure simultaneous selection procedure. The second one. for data sets with mam predictors, a backward elimination procedure for simultaneous selection is also presented. The procedures are based on Bayesian model selection criteria. An example is given to illustrate the methodology. |
| |
| Keywords: | Bayesian criterion Box-Cox transformation Calibration number Comparison score Backwards selection |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
