Development of geospatial analysis method to detect outlying data points

摘要

Subsurface investigation results always reflect a level of soil uncertainty, which sometimes requires statistical corrections of the data for the appropriate engineering decision. This study suggests a closed-form framework to detect the outlying data points from the whole testing results by the statistical geo-spatial information analyses with extreme value distribution-based outlier analysis and kriging-based cross-validation. The spatial distribution of soil layer thickness in the area of Yeouido, Seoul, South Korea is analyzed to identify the outlying data points out of the given boring data