Outlier Detection Using Density-Based Clustering: A Comparative Study of LOF and ARDOD Algorithms

Abstract

Outlier detection (OD) plays an important role in areas such as fraud detection, network security, and so on. In addition, traditional OD methods were limited to detecting a single type of anomaly: local, global, or clustered anomalies. For denser regions, extraction of outliers may turn out to be a challenging task. In order to address these issues, the paradigm of density-based outlier detection (DBOD) came into existence. DBOD mainly leverages the ideas of local outlier factor (LOF), the reachability distance, and density. The LOFbased OD approach is examined in detail to provide a clearer understanding. It is then compared with another widely used OD method: Adaptive Radius Density-Based Outlier Detection (ARDOD).