文章基本信息

标题：Comparison Of Robust Estimator In Case Of Outliers
本地全文：下载
作者：Shahbaz Nawaz ; Naeem Shahzad ; Tayyab Raza Fraz 等
期刊名称：Webology
印刷版ISSN：1735-188X
出版年度：2022
卷号：19
期号：2
页码：8994-9001
语种：English
出版社：University of Tehran
摘要：Outliers is a big problem in real life data analysis. In case of outliers, simple linear regression cannot perform well. For this problem, robust type of estimators are present. In this study, a simulation study is done from normal distribution having a sample size of 2500. Outliers with different percentages are generated to observe the efficiency of the robust type estimators. Three types of maximum likelihood (M) and modified maximum likelihood (MM) are used for the purpose of analysis. The efficiency is observed for each estimator and the coefficients are noted. The comparison is made with ordinary least square (OLS) in case of no outliers and for different percentages of outliers in the dataset. The results are observed in each case. Overall the Huber M showed the better efficiency than other estimators in the generated scenarios.
关键词：robust regression;outliers;least square;simulation. M estimators