Review of Applicable Outlier Detection Methods to Treat Geomechanical Data

The reliability of geomechanical models and engineering designs depend heavily on high-quality data. In projects, collecting analyzing laboratory data is crucial in characterizing the mechanical properties soils rocks. However, insufficient lab or underestimating treatment can lead to unreliable being used design stage, causing safety hazards, delays, failures. Hence, detecting outliers extreme values significant for ensuring accurate analysis. This study reviews categorizes applicable outlier detection methods into fence labeling statistical tests. Using real data, applicability these was examined based four elements: distribution, sensitivity values, sample size, skewness. results indicated that tests were less effective than due limitations handling skewed small sizes. Thus, best method should consider this matter. Fence methods, specifically, medcouple boxplot semi-interquartile range rule, identified as most but may necessitate more advanced techniques. Moreover, Tukey’s found unsuitable negative confidence intervals conflicted with principles.