Maximum product spacings method for the estimation of parameters of linear regression

Abstract

© Published under licence by IOP Publishing Ltd. Maximum product of spacing (MPS) estimator, which is a general method for estimating parameters from observations with continuous univariate distributions, is considered as an alternative approach in linear regression modelling. We describe the basic idea of the maximum spacings estimator and apply to the linear regression problem. Moreover, we conduct a simulation and experiment study to make the comparison between MPS method and maximum likelihood estimator under various distribution assumptions. Finally, a real data set has been implemented to illustrate the performance of this estimator.