Detecting Influential observations in Two-Parameter Liu-Ridge Estimator

Journal of Data Science

article

Adewale F. Lukman¹  Kayode Ayinde² 

[1]Department of Mathematics, Landmark University

[2]Department of Statistics, Federal University of Technology

关键词: Influential Statistics;    Multicollinearity;    Diagnostic Measures;   

DOI  :  10.6339/JDS.201804_16(2).0001

学科分类：土木及结构工程学

来源: JDS

【 摘 要 】

Influential observations do posed a major threat on the performance of regression model. Different influential statistics including Cook’s Distance and DFFITS have been introduced in literatures using Ordinary Least Squares (OLS). The efficiency of these measures will be affected with the presence of multicollinearity in linear regression. However, both problems can jointly exist in a regression model. New diagnostic measures based on the Two-Parameter Liu-Ridge Estimator (TPE) defined by Ozkale and Kaciranlar (2007) was proposed as alternatives to the existing ones. Approximate deletion formulas for the detection of influential cases for TPE are proposed. Finally, the diagnostic measures are illustrated with two real life dataset.

【 授权许可】

CC BY