Inferential Study using Neutrosophic Imputation Techniques in Two-Phase Sampling

Abstract

 In the presence of supplementary information, classical statistical methodologies typically rely on
 precise data to obtain efficient estimators of the population mean. However, the presence of outliers poses
 substantial challenges to these traditional techniques, which are highly sensitive to data accuracy and auxiliary
 inputs. In contrast, neutrosophic statistics offers a more flexible and robust framework capable of handling
 imprecise and uncertain data, thus providing an advantageous alternative to classical methods. In the present
 research, we modify the study of Gohain et al. ( [1]) by adapting his estimators and proposing a new family
 of neutrosophic exponential–logarithmic-type estimators in the neutrosophic setup with a view to enhancing
 estimation accuracy in two-phase sampling. Specifically, three imputation methods are developed, each ac
companied by their respective point estimators. Theoretical properties, including bias and expressions for the
 minimum mean square error (MSE), are derived under large sample approximations. By integrating exist
ing imputation methodologies within the neutrosophic framework, this research enhances the scope of current
 statistical inference techniques and underscores the adaptability of the proposed approach. A comparative eval
uation is conducted to assess the relative efficiency of the proposed imputation procedures against alternative
 methods considered in this study. The empirical analysis, based on real-world datasets, demonstrates the supe
rior performance of the proposed estimators. Furthermore, the findings are corroborated through an extensive
 simulation study, thereby reinforcing the validity and practical relevance of the proposed methodology

DOI: 10.5281/zenodo.17102669