Review and improvement on several typical regression diagnostics
暂无分享,去创建一个
Review is given on two kinds of regression diagnostics approaches including eight series of algorithms. Based on theoretical analysis and simulation, some shortcomings are pointed out. Three improvements are depicted, and a new detection algorithm is built, which is based on fault-tolerant identification and jackknife algorithms. Simulation shows that this new method is very efficient.
[1] N. Draper,et al. Applied Regression Analysis , 1966 .
[2] S. Weisberg,et al. Characterizations of an Empirical Influence Function for Detecting Influential Cases in Regression , 1980 .
[3] F. J. Anscombe,et al. The Examination and Analysis of Residuals , 1963 .