Methods used for identifying the influential observations in regression analysis and their comparison

---

Thesis Type: Postgraduate

Institution Of The Thesis: Gazi University, Fen Bilimleri Enstitüsü, İSTATİSTİK ANABİLİM DALI, Turkey

Approval Date: 2014

Thesis Language: Turkish

Student: CAN DARICA

Supervisor: HAMZA GAMGAM

Open Archive Collection: AVESIS Open Access Collection

Abstract:

Influential observation and/or observations in regression analysis may have significant impact on the alteration of the MLS (Method of Least Squares) estimates. Since these effects upon the estimates reduce the accuracy and efficiency of the model, detecting the influential observation and/or observations in a dataset matters to a great extent. In this study Cook Distance, DFFITS, S and M diagnostic statistics have been reviewed. The mentioned identification statistics have been compared with regards to their extent to which they can assign influential observation and/or observations in a dataset consisting of one or two influential observations.