March 3, 2026Open Access

Improvement of the Population Mean Using Cosine Robust Regression Estimator

Key Points

Robust regression techniques better estimate the finite population mean, enhancing accuracy.
The novel cosine robust estimators demonstrate lower Mean Square Error compared to Ordinary Least Squares.
Analysis used simulated and actual data to validate theoretical properties of estimators.
Findings highlight the advantages of using robust methods over conventional techniques in practical applications.

Abstract

This paper introduces a novel family of cos-robust regression-type estimators along with special members to estimate the finite population mean under SRSWOR. The new family of estimators is produced by hybridizing the auxiliary information with the cos function. To reduce the impact of outliers, various robust regression techniques, namely Huber’s M- estimation, Mallows’ GM-estimation, Schweppe’s GM-estimation, and SIS GM-estimation, are employed and theoretically compared with the Ordinary Least Squares (OLS) method. Simulated and actual data are used to generate and validate theoretical properties, such as bias and Mean Square Error (MSE). According to the findings, the robust estimators perform better than the conventional OLS approach in terms of MSE and PRE. Mathematics Subject Classification (2020). 62D05,62G05, 62G35, 62J05

Improvement of the Population Mean Using Cosine Robust Regression Estimator

Key Points

Abstract

Cite This Study