Hayder Raaid Taliba, Hassan Hopooq Razaqb, Sarah Adel Madhiomc
a University of Sumer, College of Administration and Economics, Iraq.
b University of Thi-Qar, Faculty of Administration and Economics, Iraq.
c Middle Technical University, Suwaira Technical Institute, Iraq.
The issue of incomplete data presents one of the major challenges in statistical analysis, as missing observations can significantly reduce the accuracy and reliability of estimation results. This study addresses this problem by comparing several classical and robust estimation methods used for analyzing regression model parameters in the presence of incomplete data.
The methods under consideration include R-Estimators, L-Estimators, the Expectation–Maximization (EM) algorithm, and W-Estimators. The performance of these methods was evaluated under both normal data distributions and simulated missing data patterns. The efficiency and robustness of each approach were examined, and their effectiveness compared against the Maximum Likelihood Estimation (MLE) method.
The study further explores the practical implications of these estimation techniques using real-world economic data representing demand, government consumption, and consumer spending for the period 2000–2022. The findings suggest that robust estimation methods outperform classical ones in terms of accuracy and stability when dealing with incomplete datasets.
Keywords: R-Estimators; L-Estimators; EM algorithm; W-Estimators; Incomplete data; Robust methods.
Talib, H. R., Razaq, H. H., & Madhioom, S. A. (2023). Comparison between classical and robust estimation methods for regression model parameters in the case of incomplete data. Advances in the Theory of Nonlinear Analysis and Its Application, 7(4), 17–34.