New Stochastic Restricted Estimator: A Step Towards Improved Modeling in Linear Regression

Here, we propose a new convex linear combination estimator, called the New Mixed Estimator (NME), for multiple linear regression models with stochastic linear constraints on unknown parameters and multicollinearity between explanatory variables. The Ordinary Mixed Estimator (OME) and the Biased Stochastic Restricted Liu-type Estimator (SRLIT), two well-known estimators, are integrated to create the NME. Using Mean Square Error (MSE) as the main performance parameter, we examine the statistical characteristics of the NME and theoretically compare it to both OME and SRLIT to show its superiority. According to our research, the NME routinely performs better than the OME and SRLIT in terms of overall efficacy and statistical characteristics. Additionally, we provide a numerical example to clearly support these findings.