On Composite Estimation Utilizing Regression and Classification Method

• Autorzy: Wojciech Gamrot
• Języki publikacji: EN

Abstrakty

EN

Several estimation procedure have been developed to compensate for the deterioration in properties of parameter estimates resulting from sample data incompleteness. Most of them make use of available auxiliary data following one of two general approaches. The first approach relies on dependencies between auxiliary variables and the variable under study. This usually leads to the construction of various ratio and regression estimators. The second approach explores dependencies between auxiliary variables and response behaviour of population units. This provides motivation to a broad range of methods such as weighting adjustments and classification estimators. In this paper a composite estimator of the population mean incorporating both approaches is considered. It is constructed as a combination of the well-known regression estimator and a classification estimator utilizing Bayesian quadratic discrimination function. The weights of the combination reflect the regression model's goodness of fit and the classification quality. Hence, greater weight is assigned to the estimator for which available observations of auxiliary variables are more useful. Simulation results exposing its properties are presented in the paper. (original abstract)

Słowa kluczowe

EN

Estimation; Estimation methods

PL

Estymacja; Metody estymacji

• Czasopismo: Studia Ekonomiczne / Akademia Ekonomiczna w Katowicach
• Rocznik: 2009
• Numer: nr 53 Metody wnioskowania statystycznego w badaniach ekonomicznych
• Strony: 9--18

Twórcy

autor

Wojciech Gamrot

• The Karol Adamiecki University of Economics in Katowice, Poland

Bibliografia

• Duda R.O., Hart P.E. and Stork D.G. (2001): Pattern Classitication. Wiley & Sons. Inc., New York.
• Gamrot W. (2003a): On Applicalion of Some Discrimination Methods to Mean Value Estimation in the Presence ot Nonresponse. In: J. Wywial (ed.): Metoda reprezentacyjna w Badaniach Ekonomiczno-Społecznych. AE. Katowice. pp. 37-50.
• Gamrot W. (2003b): A Monte Carlo Comparison of Some Two-Phase Sampling. Strategies Utilizing Discrimination Methods in the Presence of Nonresponse. "Zeszyty Naukowe". No. 29, University of Economics. Katowice, pp. 41-54.
• Hansen M.H.. Hurwitz W.N. (1949): The Problem of Nonresponse in Sample Surveys. "Journal of the American Statistical Society". No. 41, pp. 517-529.
• Sarndal C.E., Swensson B. and Wretman J. (1997): Model Assisted Survey Sampling. Springer Verlag, New York.
• Wywiał J. (2001): On Estimation of Population Mean in the Case When Nonrespondents Are Present. "Prace Naukowe". 8, 906, AE. Wrocław, pp. 13-21.