COMPARISON OF SOME ROBUST REGRESSION METHODS IN CASE OF OUTLIER


Abstract views: 87 / PDF downloads: 240

Authors

  • Öznur İŞÇİ GÜNERİ Muğla Sıtkı Koçman Üniversitesi, Fen Fakültesi, İstatistik Bölümü
  • Aynur İNCEKIRIK Manisa Celal Bayar Üniversitesi, İktisadi ve İdari Bilimler Fakültesi, Ekonometri Bölümü
  • Burcu DURMUŞ Muğla Sıtkı Koçman Üniversitesi, Fen Fakültesi, İstatistik Bölümü

DOI:

https://doi.org/10.51296/newera.133

Keywords:

Quantile Regression, LAD Regression, M Regression, MM Regression

Abstract

The presence of outliers or observations in the data set in the studies can significantly affect the statistical analysis results and modeling. The least squares method which is sensitive to outliers can also give misleading results when the assumptions are not met. In this case robust regression methods which are presented as an alternative to multiple linear regression, are used. In this study a sample data set was taken and a study was conducted to investigate how much the regression estimators could explain the data set in case of outliers within the observation points. For this purpose, in case of outliers, quantile regression method from robust regression methods, smallest absolute deviations (LAD), which is a special case of quantile regression method, commonly used M estimators among robust estimators, with S and MM estimators with high breakpoints were used.

Downloads

Published

2021-10-15

How to Cite

İŞÇİ GÜNERİ, Öznur, İNCEKIRIK, A., & DURMUŞ, B. (2021). COMPARISON OF SOME ROBUST REGRESSION METHODS IN CASE OF OUTLIER. NEW ERA INTERNATIONAL JOURNAL OF INTERDISCIPLINARY SOCIAL RESEARCHES, 6(11), 33–51. https://doi.org/10.51296/newera.133