PENDEKATAN REGRESI ROBUST DENGAN FUNGSI PEMBOBOT BISQUARE TUKEY PADA ESTIMASI-M DAN ESTIMASI-S

Ana Nurbaroqah (Jurusan Matematika, Universitas Jenderal Soedirman)
Budi Pratikno (Universitas Jenderal Soedirman)
Supriyanto Supriyanto (Jurusan Matematika, Universitas Jenderal Soedirman)

Jurnal Ilmiah Matematika dan Pendidikan Matematika (JMP)
Vol 14 No 1 (2022): Jurnal Ilmiah Matematika dan Pendidikan Matematika

Least Square Method is one of methods for estimating of parameters of regression model. Model of least square methods is not valid if there are some disobeydiance in classical assumptions, for example, there are outliers. To resolve the problem, robust regression method is usually used. The method is used because it can detect the outliers and give stable results. In this research, data used is data for human development index of districts in Central Java from 2019 to 2020. Estimation for robust regression method chosen is estimation-M and estimation-s with Tukey Bisquare as a weight function. Criterions for choosing the best model are based on adjusted R-Squared value and mean square error (MSE). The result shows that robust regression model with estimation-M is a better model since it has adjusted R-Squared value tending to one and the least MSE.