A Study on Different Methods of Outlier Detection Algorithms in Data Mining

Abstract

In the modern world, data are available widely, and it is essential to transform such data into useful knowledge and information. Data mining has attracted considerable awareness in the field of information and its community. Dataset is a collection of significant objects which do not belong to the same category. Some objects differ slightly from other regular objects are identified as outliers. Detecting outliers are notable because its presence slows down the system performance. Most methods of data mining dismiss outliers as noise or exceptions. However, rare events can be more attractive in some applications, such as fraud detection than frequent events. Outlier analysis is also known as outlier mining. Many fields such as marketing, sales, production, fraudulent identification, customer retention, and scientific research, use the acquired data. Rough sets are used to handle uncertain and vague data present in the real world. The discussion of rough classification, clustering, and different outlier detection methods are carried out in detail with suitable algorithms and examples. This survey provides an overview of outliers and existing outliers by classifying them into different dimensions. Wine dataset from the UCI repository has been taken to prove the performance of the rough set based entropy measure with weighted density value over existing methods.