Review of Outlier Detection Methods

Rasoul Kiani1 Mojtaba Montazeri2

1) Department of Computer Engineering, Fars Science and Research Branch, Islamic Azad University, Marvdasht,Iran Email:
2) Department of Computer Engineering, Fars Science and Research Branch, Islamic Azad University, Marvdasht,Iran Email:

IOI Code : XYEZ-AHEEE
Publication : International Conference On Research Science And Technology(rstconf.com)
Abstract :
    The first purpose of preprocessing phase in data mining is eliminating data problems in the problem that is being studied. Outlier detection and removing in data cleansing process are the actions that can improve the data quality. Different ways of outlier detection in intrusion detection, fraud detection, medical informatics, damage detection, image processing and so on, make the necessity of available study methods more clear. In this paper after reviewing outlier concepts such as key components and different type of outliers, primary algorithms according to the application domains and anomaly are studied and compared and also new high usage methods are studied in detail. The aim of this research is presenting a coherent and applicable study according to the current methods in outlier detection.

Keywords : outlier detection anomaly detection