Regression Model Estimation Using Least Absolute Deviations , Least Squares Deviations and Minimax Absolute Deviations Criteria
نویسندگان
چکیده
Regression models and their statistical analyses is the most important tool used by scientists in data analyses especially for modeling the relationship among random variables and making predictions with higher accuracy. A fundamental problem in the theory of errors, which has drawn attention of leading mathematicians and scientists since past few centuries, was that of fitting functions. For the pioneering work to develop procedures for fitting functions, in the eighteenth century, we refer to the research by Mayer, Boscovich, Laplace, Legendre, Simpson, Gauss, and Euler. They worked on the methods of least absolute deviations, least squares deviations and minimax absolute deviations. Today’s widely celebrated procedure of the method of least squares for function fitting is credited to the published works of Legendre and Gauss [1][2]. The least square estimates are best linear unbiased estimates and are optimal under assumptions that the errors follow normal distributions, are free of large size outliers and satisfy the GaussMarkov assumptions. However, the least squares based models in practice may fail to provide optimal results in non-Gaussian situations especially when the errors follow distributions with the fat tails. In this paper, we will present an overview of some important works in fitting linear relationship. We will also statistical properties of the Lp -norm based model estimation, measures of the model adequacy and also future research questions of interest. Keywords— Regression model, Least squares estimates, Least absolute deviations, Minimax absolute deviations, Lp-norm. I. FITTING FUNCTIONS: INTRODUCTION AND REVIEW standard linear regression model expressing the linear relationship between the study variable and explanatory variables is defined by linear equations:
منابع مشابه
Least Absolute Value vs. Least Squares Estimation and Inference Procedures in Regression Models with Asymmetric Error Distributions
متن کامل
A Note on Alternative Matrix Entry Estimation Techniques
Estimation of trip tables and other matrices that are subject to constraints is a common practical problem. This note reviews four common estimation methods: (1) minimization of the sum of absolute deviations, (2) the biproportional technique, (3) information minimization and (4) constrained generalized least squares (CGLS) regression. A small example illustrates their application. Computationa...
متن کاملLeast Absolute Deviations Estimation for the Accelerated Failure Time Model
The accelerated failure time (AFT) model assumes a linear relationship between the event time and the covariates. We propose a robust weighted least-absolute-deviations (LAD) method for estimation in the AFT model with right-censored data. This method uses the Kaplan-Meier weights in the LAD objective function to account for censoring. We show that the proposed estimator is root-n consistent an...
متن کاملLeast squares and shrinkage estimation under bimonotonicity constraints
In this paper we describe active set type algorithms for minimization of a smooth function under general order constraints, an important case being functions on the set of bimonotone r × s matrices. These algorithms can be used, for instance, to estimate a bimonotone regression function via least squares or (a smooth approximation of) least absolute deviations. Another application is shrinkage ...
متن کاملLeast Absolute Deviation Estimation of Linear Econometric Models : A Literature Review
I. Introduction: The Least Squares method of estimation of parameters of linear (regression) models performs well provided that the residuals (disturbances or errors) are well behaved (preferably normally or near-normally distributed and not infested with large size outliers) and follow Gauss-Markov assumptions. However, models with the disturbances that are prominently non-normally distributed...
متن کامل