Estimation and Simulation of Parameters in Beta Ridge Regression

In the analysis of regression problems and especially statistical modeling in economic data, psychology, social sciences, vital,engineering, etc., we face the problem of collinearity among predictor variables and autocorrelation in errors. In such cases, theordinary least squares estimator leads to imprecise estimates in terms of magnitude and sign and results in wide confidence intervalsfor the parameters. The purpose of this article is to investigate multiple collinearity problems and overcome it by introducing thegeneralized limited differential ridge regression estimator and comparing it with the ordinary generalized least squares estimator.This estimator is a generalized bounded differential estimator that is produced by minimizing the sum of the square powers of theresiduals and considering a spherical restriction on the parameter space. The hazard function of the proposed estimators iscalculated from the balanced language function. Finally, the performance of the new processor is evaluated using simulated data.

Keywords:

Ridge Estimator , Beta Regression , Balanced Language Function , Linear Constraints