Studi Penaksiran Parameter Model Regresi Linier Berganda Berdasarkan Metode OLS, Jackknife dan Bootstrap

Abstract

This study aims to estimate the parameters of the Multiple Linear Regression model. The methods used to estimate parameters are the OLS, Bootstrap and Jackknife methods. The results of these estimates will be compared to find out the best method used. In this study B was used in the amount of 10, 100, 500, 1000, 5000, 10000, and 50000. The sample reduction (d) used in this study was 1, 2, and 3. After simulating the data, it was found that the larger the size of B does not guarantee a decrease in MSE, RMSE and SE. This happened because of the residual sampling of Bootstrap and Jackknife samples which were different for each sampling. From the analysis above, it shows that in the data that there are few outliers, the Bootstrap and Jackknife methods are not always better at estimating parameters compared to the OLS method. However, for data that has large outliers, the best method to use is the Jackknife method. This is opened by the RMSE, MSE and SE values which are smaller than the OLS method and the Residual Bootstrap method.