RESIDUAL BOOTSTRAP RESAMPLING METHOD FOR MULTIPLE LINEAR REGRESSION MODEL PARAMETER ESTIMATION

Abstract

The Ordinary Least Square (OLS) method is a standard method for estimating parameter values for simple linear and multiple linear regression models.Bootstrap resampling method is divided into 2, the residual bootstrap resampling method and bootstrap pairs.This study aims to estimate the value of the regression parameters using the residual bootstrap resampling method to analyse the effect of school participation rates, percentage of junior high school graduates, households with access to clean water, labour force participation rates, open unemployment rates, and GRDP on the Human Development Index in Central Java Province 2018.The results of the analysis using the standard method cannot be used because the assumptions are not met.As an alternative the residual bootstrap resampling method is used.Based on the analysis carried out, obtained a residual bootstrap resampling method that has a smaller standard error is se 〖β _0〗^*=4.84324491,〖〖seβ 〗_1〗^*=0.04579879,〖se β _2〗^*=0.05217101.From the results of comparison using MSE it is known that the smallest MSE value using B = 2000.Therefore, it can be concluded that the resampling method of residual bootstrap is the right and best by using sample size 35 with B = 2000.