| Abstrakt: |
In regression analysis, the ordinary least squares (OLS) technique is a fundamental tool that is frequently used to model relationships between variables. If the data set contains an outlier, this technique could produce inaccurate outcomes. Outliers, or data points that are significantly different from the rest of the data, can lead to incorrect statistical interpretation. Systematically addressing outliers improves data precision and efficiency prior to drawing conclusions. Outliers may be misleading and cause statistical findings to be interpreted incorrectly. Prior to drawing conclusions from the data, it can be improved to acquirerelevant information by dealing with outliers effectively. The normality assumption is an essential statistical presumption that issusceptible to the effect of outliers in the context of traditional least squares regression. On the other hand, robust regression is a method used to estimate the regression coefficients in a linear regression model wherein the data contains outliers in an effort to improve the accuracy of the estimations. In the context of sample surveys, robust regression methods have been commonly suggested as useful alternatives to ordinary least squares for dealing with outliers. A novel exponential estimator for estimating the finite population mean is presented in this paper, which uses robust regression techniques while taking advantage of the known parameters of an auxiliary variable associated with the study variable. Assuming that the data contains outliers and that outliers have no effect on these estimators, as we are using the robust regression method, which is obtained by Huber M-estimation, our suggested estimators' mean squared errors are determined theoretically and compared to the ordinary least squares techniques. The numerical illustration and simulation results demonstrate that, when using the robust regression method, the suggested estimators perform well in the presence of outliers. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink