ESTIMASI METHOD OF MOMENT DENGAN FUNGSI PEMBOBOT HUBER PADA PEMODELAN DATA TERIDENTIFIKASI PENCILAN

Abstract

The purpose of this study is to obtain a robust estimator using the moment method using Huber weighting in multiple linear regression modeling on the identified data and compare the results of the analysis between weighted Bisquare Tukey and Huber weights. The usual methods used to estimate regression parameters include the Least Square Method (MKT). Robust regression is a method that can generate a robust parameter estimator of the outliers. A robust estimator is relatively unaffected to large changes in small pieces of data or small changes in large parts of the data. The MM estimation is a good method to cope with the results of robust estimators (robust) with high breakdown points and high efficiency and involves multiple weights, the weighted Bisquare Tukey and Huber weights. The result of the analysis shows that the data is identified by the method of moment (MM) using weighted Bisquare Tukey more robust than using Huber weights it is aimed with the coefficient of determinant value (R2) and the middle quadratic error value (KTG) gives result Which is better and more efficient.

Keywords: Least square Method, Evalengism, Weighing Bisquare Tukey,
Weighted Huber, Robust Regression with MM Estimation