Neutrosophic Method for Identifying Extreme Values in Imprecise Data Using Median Absolute Deviation

The traditional calculation of Z-scores for outlier detection is highly sensitive to extreme data points, making it unsuitable under conditions of uncertainty. In this study, we propose a novel approach to modify the Z-score method using neutrosophic statistics. Key statistical measures, including the median, neutrosophic standard deviation, and median absolute deviation, will be computed based on neutrosophic random variables. An extensive simulation study will evaluate the impact of varying uncertainty levels on the adaptation of Z-scores for outlier detection and their effectiveness in identifying outliers. Comparative analysis of Z-scores derived from different methods will also be performed. The proposed methodology will be applied to neutrosophic GG25 gray cast iron data, demonstrating its practical utility. We hypothesize that uncertainty levels will significantly affect Z-score computations and, consequently, outlier detection in the dataset.