BIASED ESTIMATOR BASED ON DIAGNOSIS AND MEASURE OF MULTICOLLINEARITY
Zhang Lei 1) ;Gu Yongwei 1,2) ;Gui Qingming 1) ; and Ma Chaozhong 1)
1)Institute of Science, Information Engineering University, Zhengzhou 4500012)Institute of Surveying and Mapping, Information Engineering University, Zhengzhou 450052
Abstract:On the basis of diagnosis and measure of multicollinearity, a new biased estimator of unknown parameters called partial ridge (PR) estimator is proposed for GaussMarkov model. Its properties are discussed, and some important conclusions are drawn. The determination of biased parameter in the PR estimator is discussed too. Both theoretical and computational results demonstrate that the PR estimator is a effective biased estimator for overcoming the effect of multicollinearity and is superior to the ordinary ridge estimator.