Mengatasi Outlier dengan Metode Least Trimmed Squares (Lts) pada Regresi Robust

Abstract

This study is to get a regression equation better than regression equation before for data have outlier. First, check outlier at data, with grafic and looking for residu studenization, leverage value, DfFitS, DfBETAS(s) and Cook’s Distance. And then searching regression equation with Least Trimmed Squares (LTS) method at robust regression, that is with get total of sum minimum kuadrat residu with coverage measured. It will get regression equation with LTS method better than equation before with OLS because LTS can make outlier influence be smaller than before for data.