ESTIMASI PARAMETER MODEL REGRESI LINIER DENGAN PENDEKATAN METODE BAYESIAN

Abstract

The method used to estimate the regression parameters in this study is the Bayes method. The steps in estimating the parameters of the linear regression model using the Bayes method are to determine the Likelihood function from the normal distribution density function, determine the Prior Non-Informative distribution from a normal distribution, then look for the Posterior distribution by switching the Prior distribution with the Likelihood function.Writing this thesis aims to determine the estimation of the parameters of the linear regression model with the Bayesian method approach. From the regression model generated by the OLS method, it was identified that there was one Outlier data. Outlier is a factor that influences parameter estimation in linear regression model. A model will be shown in an example data case by comparing the results of the Ordinary Least Square (OLS) method using R and the results of the Bayesian method using WinBUGS. It can be seen that the MSE value obtained from the Bayesian Method estimation is smaller than the MSE value obtained from the OLS Method estimate. The Mean Square Error (MSE) value obtained from the estimation of the OLS Method is 0.8469 while the Mean Square Error (MSE) value obtained from the estimation of the Bayesian MCMC Method is 0.3723. This shows that the Bayesian MCMC method is much better than the OLS method.