A Survey on Mixed-Attribute Outlier Detection Methods

In the data era, outlier detection methods play an important role. The existence of outliers can provide clues to the discovery of new things, irregularities in a system, or illegal intruders. Based on the data, outlier detection methods can be classified into numerical, categorical, or mixed-attribute data. However, the study of the outlier detection methods is generally conducted for numerical data. Meanwhile, many real-life facts are presented in mixed-attribute data. In this paper, the researcher presents a survey of outlier detection methods for mixed-attribute data. The methods are classified into four types, namely, categorized, enumerated, combined, and mixed outlier detection methods for mixed-attribute data. Through this classification, the methods can be easily analyzed and improved by applying appropriate functions. © 2019 Bina Nusantara University. All right reserved.