Simulation Study using Average Difference Algorithm to Analyze the Outlierness Degree of Spatial Observations

Abstract

Attribute values are the main elements in calculating degree of outlierness of spatial objects. The problem arises when the spatial outliers with extreme values are the nearest neighbors of a central object. In this study, several scenarios are simulated to verify the effect of spatial outliers’ extreme values to the degree of outlierness of its nearest neighbors, based on Average Difference Algorithm. The results confirmed the effect can lead to falsely detected spatial outliers. The algorithm detect the true spatial outliers correctly if their values are three sigma away from the mean attribute values of its nearest neighbors.