1
Master's student, Department of Statistics, Faculty of Mathematics, Yasuj University, Yasuj, Iran
2
Department of Statistics, Faculty of Mathematics, Yasuj University, Yasuj, Iran
Abstract
The traditional linear regression model is represented as Y = Xβ + ε, with the estimated parameter β calculated as ˆ β = (X′X)−1X′Y. However, when implementing this estimator in practical applications, several issues may arise, such as vari able selection, collinearity, high dimensionality, dimension reduction, and measurement error, which can make it challenging to use the above estimator. The primary problem in most of these cases is the singularity of the matrix X′X. A variety of solutions have been proposed to address these problems. In this article, we review these issues and present a comprehensive set of commonsolutions, as well as some advanced and less commonly used methods, that have the potential to address these problems in an intelligent manner.
