作者: José García , Román Salmerón , Catalina García , María del Mar López Martín
DOI: 10.1111/INSR.12099
关键词:
摘要: Summary Ridge estimation (RE) is an alternative method to ordinary least squares when there exists a collinearity problem in linear regression model. The variance inflator factor (VIF) applied test if the original model and also necessary after applying ridge estimate check chosen value for parameter k has mitigated problem. This paper shows that application of data working with leads non-monotone VIF values. Garcia et al. (2014) showed some problems traditional used RE. We propose augmented VIF, VIFR(j,k), associated RE, which obtained by standardizing before augmenting VIFR(j,k) will coincide estimator = 0. very desirable properties being continuous, monotone higher than one.