NON-PARAMETRIC REGRESSION ESTIMATION FOR DATA WITH EQUAL VALUES

Abstract

Parametric regression analysis depends on some assumptions. One of the most important of assumption is that the type of relationship between dependent and independent variable or variables is known. Under such circumstances, in order to make better assumptions, regression methods which enable flexibility in the linearity assumption of the parametric regression are needed. These methods are nonparametric methods known as semi parametric regression methods. Estimation of parameters in a parametric regression which has independent variables of different values has been studied extensively in literature. Sometimes, one or more observation series of independent variable values can be equal while dependent variable values are different. This study offers a new method for the estimation of regression parameters under such data. Proposed method and other nonparametric methods such as Theil, Mood-Brown, Hodges- Lehmann methods and OLS method were compared with the sample data and the results were evaluated.