A Comprehensive Study on Outlier Detection in Data Mining

Abstract

The paper presents a survey on the literature of outliers and data mining. The prime focus of the paper is to deliver an outline of outlier with various approaches for its detection. Outliers are the data points which are partially or totally diverge from the residue data set. They can be considered as those data objects which cannot be fitted in any cluster. Outliers can be different from its neighboring data points only or from complete data set. It is a necessary task to identify and detect outliers from the dataset as their presence effect the preciseness of the outcome. Outliers can exist in any kind of data varies from low-dimensional to high-dimensional data set. The detection of an outlier requires some precise mathematical calculations, appropriate domain knowledge, and statistical calculations which are presented in the paper. The paper presents, the significant characteristics of the outliers.