Performance Evaluation of Outlier-Detection Algorithms using Various Parameters

Data mining is one of the highly researched areas in the present world due to the tremendous increase of data. With the increase in the data dimensionality in various domains, effectiveness and efficiency of traditional mining techniques are now a great challenge. So there arises a need to process large datasets, which produces several challenges for researchers. Many outlier-detection methods are used for detecting outliers. Outlier detection is the process of finding out the data objects having a different behaviour from expectations. Various algorithms have been proposed till date for the detection of the outliers. Data clustering proposes the division of database into groups of objects, depending upon the similarity among the objects. This paper covers a study of various clustering-based outlier-detection algorithms like K-means, Fuzzy K-means, Iterative K-means and depth-based clustering outlier-detection algorithm. Comparison study of these clustering-based outlier-detection methods is done to find out which of the outlier-detection algorithms are more applicable on high-dimensional data.