| 关于多重共线性检验方法的研究 |
| |
| 引用本文： | 赵松山,白雪梅.关于多重共线性检验方法的研究[J].山东工商学院学报,2001,15(4):296-300. |
| |
| 作者姓名： | 赵松山 白雪梅 |
| |
| 作者单位： | 1. 东北财经大学,数量经济系,辽宁,大连,116025
2. 东北财经大学,统计系,辽宁,大连,116025 |
| |
| 摘 要： | 首先指出多重共线性是指解释变量观测值矩阵 X的秩小于待估参数个数 ,以及存在多重共线性带来的严重影响 ;其次介绍了 8种检验多重共线性的方法 ,并加以简要评述 ;最后特别强调近几年对多重共线性研究的成果 ,介绍了有别于从变量出发的数据影响点的新思路 ,给出两种多重共线性影响点的诊断方法——特征分析法及主成分诊断法。
|
| 关 键 词： | 多重共线性 判定系数 偏相关系数 病态指数 影响点 |
| 文章编号： | 1006-6160(2001)04-0296-05 |
| 修稿时间： | 2001年3月20日 |
Research on the Test for Multi-co linearity |
| |
| ZHAO Song-shan,Bai Xue-mei.Research on the Test for Multi-co linearity[J].Journal of Shandong Institute of Business and Technology,2001,15(4):296-300. |
| |
| Authors: | ZHAO Song-shan Bai Xue-mei |
| |
| Institution: | ZHAO Song-shan 1,Bai Xue-mei 2 |
| |
| Abstract: | The paper points out the definition of multi-co linearity. The rank of explanatory variables' observation matrix is less than the sum number of explanatory variables plus one. Besides, effects arising from the existence of multi-co linearity is addressed. Eight approaches to test multi-co linearity are introduced and assessed. Current research resulting from this issue is emphasized. The new idea-data-influence-points instead of the based on variables is illustrated. Moreover, two testing models for influence points are presented-character analysis and principle component diagnosis. |
| |
| Keywords: | multi-co linearity coefficient of determination partial correlation coefficient ill-conditioned index influence point |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
