| 一种新的回归变量选择方法 |
| |
| 引用本文： | 李世玲,张富堂.一种新的回归变量选择方法[J].计算机应用与软件,2006,23(5):113-115. |
| |
| 作者姓名： | 李世玲 张富堂 |
| |
| 作者单位： | 中国工程物理研究院电子工程研究所,四川,绵阳,621900 |
| |
| 摘 要： | 本文就多元线性回归模型，提出一种回归变量选择方法，其基本思想是利用矩阵的扫描运算和变量间的相关性分析，根据自变量（回归变量）与因变量之间相关性越强对因变量的解释作用越强的原则，选择回归变量。用本文提出的方法选择回归变量，具有降低回归模型复杂度、提高模型精确性和可靠性、便于工程实现的优点。仿真算例证实了该方法的有效性和可行性。
|
| 关 键 词： | 多元分析 多重相关性 扫描运算 变量选择 |
| 收稿时间： | 05 28 2004 12:00AM |
| 修稿时间： | 2004-05-28 |
A NEW METHOD FOR REGRESSION VARIABLE SELECTION |
| |
| Li Shiling,Zhang Futang.A NEW METHOD FOR REGRESSION VARIABLE SELECTION[J].Computer Applications and Software,2006,23(5):113-115. |
| |
| Authors: | Li Shiling Zhang Futang |
| |
| Affiliation: | The Institute of Electronic Engineering, China Academy of Engineering Physics, Mianyang Sichuan 621900, China |
| |
| Abstract: | By using the matrix's scanning calculation and correlation analysis among variables,a new method for regression variable selection is proposed in this paper for multiple linear regression models.Its basic principle is that,if the correlativity between independent-variables(regressive variables) is better,dependent-variables result in the better explanation function of independent-variables to dependent-variables The proposed regression variable selection method has such advantages as lower model complexity,higher precision and reliability,and better engineering practicability.The validity and feasibility of this method has been demonstrated by a simulation example. |
| |
| Keywords: | Muhianalysis Muhicollinearity Sweep operation Variable Selection |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
