Local Isolation Coefficient-Based Outlier Mining Algorithm
Bo Yu, Mingqiu Song, Leilei Wang · 2009
Outlier detection has received significant attention in many applications, such as detecting credit card fraud or network intrusions. Distance-based outlier detection is an important data mining technique that finds abnormal data objects according to some distance function. However, when this technique is applied to datasets whose density distribution is different, usually the detection efficiency and result are not perfect. With analysis of features of outliers in datasets, as the improvement of local sparsity coefficient-based (LSC) mining of outliers, we rank each point on the basis of its distance to its kth nearest neighbor and the distribution of its k nearest neighborhood. A novel outlier detecting algorithm based local isolation coefficient (LIC) is presented in this paper, which is shown better outlier mining results through the experiments.