Gazeloğlu, Cengiz Afyon Kocatepe Üniversitesi - Fen Edebiyat Fakültesi - İstatistik Bölümü, Turkey , Saraçlı, Sinan Afyon Kocatepe Üniversitesi - Fen Edebiyat Fakültesi - İstatistik Bölümü, Turkey

A simulation Study on the Perforformances of Fuzzy and Robust Fuzzy Bisectional Regression Techniques

27534

The aim of this study is comparatively examine the OLS (Ordinary Least Squares) Bisector, Fuzzy OLS Bisector, Robust OLS and Fuzzy Robust OLS techniques via Monte-Carlo Simulation study, when both the dependent and independent variables includes measurement errors in a simple linear regression analysis. In conditions, whether the interested data sets includes or not any outliers, the performance of the regression techniques are examined for different sample sizes (n=10, 50 and 100) and different distribution types (t(4), t(10),and t(30) ) according to Mean Square Error (MSE) and Relative Efficiency (R.E.) criteria. According to findings of the study, the Fuzzy Huber Bisector technique has the lowest MSE in all different distribution types and for all sample sizes for the data sets that either includes or not any outliers..

1

Type II Regression , Fuzzy Regression Robust Regression , Simulation Study

Afyon Kocatepe University Journal Of Science an‎d Engineering

13