PARAMETER ESTIMATION IN THEIL-SEN REGRESSION ANALYSIS WITH JACKKNIFE METHOD

Abstract

Regression analysis; including the cause - result relationship examines the relationship between dependent and independent variables. Estimation is one of the most widely used statistical analysis techniques. Parametric regression analysis is based on some assumptions. The most important of these assumptions, the dependent and independent variables is known of the relationship between forms. In cases not provided estimates of the assumptions made, they are not qualified to be a good estimate. In this case, in order to make better predictions that allow stretching of linearity assumption in the parametric regression methods are needed. These methods are the regression model known as nonparametric and semi-parametric regression methods. Nonparametric regression analysis, is the methods that are successful for some of the assumptions used in case of failure to provide valid parametric regression methods. Jackknife method throwing an observation at a time from the sample which statistics calculates that as the number of individuals in the sample and the effect of extreme values can be defined as a method with relieving properties. In this study, it is proposed for Theil-Sen nonparametric regression analysis using the Jackknife method. Theil-Sen mean and median of the estimates also made for proposed Jackknife method. The results obtained were compared with the jackknife method OLS and Theil-Sen methods. Theil-Sen has been determined that the jackknife results in better outcomes.