Diagnostics and plots for identifying collinearity-influential observations and for classifying multivariate outliers

Van Vuuren J.O. ; Conradie W.J. (2000)


Simple, but very effective measures and plots are proposed to detect collinearity-influential observations in multiple linear regression analysis. Unlike traditional methods that are based mostly on the eigenstructure of the regressor matrix, these methods are based solely on the general tools of regression diagnostics, namely residuals and leverage values. They lead to a better understanding of influential observations and their effect on multivariate dispersion.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/12517
This item appears in the following collections: