Latent Root Regression: Large Sample Analysis
作者:
J.W. White,
R.F. Gunst,
期刊:
Technometrics
(Taylor Available online 1979)
卷期:
Volume 21,
issue 4
页码: 481-488
ISSN:0040-1706
年代: 1979
DOI:10.1080/00401706.1979.10489818
出版商: Taylor & Francis Group
关键词: Multicollinearity;Least squares;Biased estimation
数据来源: Taylor
摘要:
Large sample properties of statistics used in latent root regression analysis are investigated by examining the matrix of correlations among the predictor and response variables as the sample size becomes infinite. The latent roots and latent vectors of the asymptotic correlation matrix are derived for specific model configurations of interest. From the study of the asymptotic latent roots and latent vectors, a new statistic is proposed for use in detecting nonpredictive multicollinearities.
点击下载:
PDF (756KB)
返 回