Performance Analysıs Of The Influence Dıstance Based On Robust Estımators For The Identıfıcatıon Of Outlıers In Lınear Regressıon Models

---

Thesis Type: Postgraduate

Institution Of The Thesis: Gazi Üniversitesi, Fen Bilimleri Enstitüsü, Turkey

Approval Date: 2020

Thesis Language: Turkish

Student: FULYA KARAKOCA

Supervisor: MELTEM EKİZ

Abstract:

In multiple linear regression analysis, it could be possible to encounter with observations that differ from the bulk of the data and are named as unusual observations in the literature. It is important to identify these observations, which are classified as outliers, influential and leverage points, in order to make accurate statistical inferences. Nurunnabi et al. (2016) suggested influence distance (ID) used to create suspicious observations. In this study, it is intended to determine unusual observations by using ID based on maximum likelihood, least median of squares, re-weighted least squares, M and S estimators. A comparison of performance was performed with a simulation study.