UTILIZATION OF DIFFERENT ROBUST REGRESSION TECHNIQUES FOR ESTIMATION OF FINITE POPULATION MEAN IN SRSWOR IN CASE OF PRESENCE OF OUTLIERS THROUGH RATIO METHOD OF ESTIMATION

Mir Subzar, Carlos N. Bouza, Amer Ibrahim Al-Omari

Resumen


Ratio type estimators are extensively used in sampling theory in order to get precise estimates of the population parameters by taking the advantage of positive (high) correlation between study and auxiliary variable than usual sample mean estimator. In this study we encountered with the problem of presence of outliers in the data and using of traditional methods usually decreases the efficiency in estimating the population parameters as these methods are sensitive to outliers. So in the present study we adapt the various robust regression techniques such as LTS, LMS, LAD, Huber M, Hampel M, Tukey M and Huber MM estimation to the ratio estimators which were suggested by Abid et al. (2016) by incorporated ancillary information using OLS method and also adapt Huber M-estimation to above estimators. Theoretically, we obtain the mean square error (MSE) for these estimators. We compared MSE values of the proposed estimators with MSE values based on Huber M which was proposed by Kadilar et al. (2007) and OLS methods. From this comparison we observe that our proposed estimators give more efficient results than both Huber M and OLS approach. These theoretical results are supported with the aid of a numerical example.
KEYWORDS: ratio type estimators, robust regression methods, ancillary information, simple random sampling, efficiency.
MSC:62D05

Texto completo:

PDF

Enlaces refback

  • No hay ningún enlace refback.