Latest revision as of 10:59, 19 March 2026

Abstract

Building efficient ratio-type estimators of population parameters, especially the mean and variance, has been a major theme in sampling theory. However, the growing frequency of dirty, inaccurate, and incomplete information still poses a threat to the credibility of the classic estimation processes, especially in environments where outliers are likely to occur. The paper derives a generalized type of neutrosophic robust ratio-type estimator and regression-type estimators that have been developed on the M-estimation platform, including Huber M-estimators and generalized M-estimators (Viz., Mallows-GM, Schweppes-GM, and SIS-GM) formulations, and also incorporating the auxiliary information of the HodgesLehmann estimator. The estimators are designed to be asymptotically efficient in clean data models and apply well to contamination and heavytailed error distributions. Through a comparative study using real-world data and a Monte Carlo simulation experiment, it is shown that the proposed estimators show better performance, numerical stability, and robustness compared to the current methods in the presence of uncertainty. The simulation results confirm their resilience to contamination and heavy-tailed distributions across varying contamination levels. An application to environmental data involving temperature measurements subject to measurement error and outliers further illustrates the practical relevance of the framework. Collectively, the empirical and simulation evidence support the applicability of the proposed methodology to industrial process analysis and environmental monitoring systems characterized by data imprecision and contamination.OPEN ACCESS Received: 16/11/2025 Accepted: 07/01/2026

Document