In quantile regression, usually the explanatory variables in medical data are highly correlated with each other. Thus, the influence of one variable can't be differentiated from the others. Also, multicollinearity generates unstable regression coefficients with undesirable large variances. It is possible that the estimated coefficients may have the incorrect signs and present problems in the interpretation. In this paper ridge regression is applied to solve the problem of multicollinearity. An optimum ridge coefficient for the ridge regression parameter can be estimated through Bayesian approach. The Bayesian approach is a method to stabilize the ridge parameter. The quantile regression is the best method to predict an extreme value. This study discusses the use of ridge regression in quantile regression with a parameter ridge. Variance inflation factor (VIF) is used to determine the best ridge coefficient. The results show that the quantile regression with ridge regression was suitable for the study of the causes of genetic anemia (Thalassemia) in children.
展开▼