Bayesian analysis of simple linear regression model: Simulation study and real life application
2013
Babar, Z.
Bayesian methods are increasingly becoming attractive to researchers in many fields. As opposed to the point estimators used by classical statistics, Bayesian statistics is concerned with generating the posterior distribution of the unknown parameters given both the data and some prior density for these parameters. As such, Bayesian statistics provides a much more complete picture of the uncertainty in the estimation of the unknown parameters, especially after the confounding effects of nuisance parameters are removed. This study is drive algebraic expressions for Bayes estimates of parameters of simple linear regression model. We are focusing the comparison of classical (Maximum Likelihood Estimates) and Bayes estimates of parameters of simple linear regression assuming the informative and non-informative (Uniform) priors. A simulation study is to be conducted to highlight properties and comparison of classical and Bayes estimates in terms of sample size, parameter size and prior information. The Minitab are used to deal with numerical composition to estimates of classical and Baye's estimates of simple linear regression models.
显示更多 [+] 显示较少 [-]