A Review of Outlier Prediction Techniques in Data Mining

S. Kannan ； K. Somasundaram

《Research Journal of Applied Sciences, Engineering and Technology》 10卷9期 (2015/07) Pp. 1021-1028

並列摘要

The main objective of this review is that to predict the outliers in data mining. In general, the data mining is a process of applying various techniques to extract useful patterns or models from the available data. It plays a vital role to choose, explore and model high dimensional data. Outlier detection refers a substantial research problem in the domain of data mining those objectives to uncover objects which exhibit significantly different, exceptional and inconsistent from rest of the data. The outlier potential sources can be noise and errors, events and malicious attack in the network. The main challenges involved in the outlier detection with high complexity, size and different types of datasets, are how to catch similar outliers as a group by using clustering-based approach. The outlier or noise available in the clustered data is accurately removed and retrieves an efficient high dimensional data. Nowadays, the classification and clustering techniques for outlier prediction are applied in various fields like bioinformatics, natural language processing, military application, geographical domains etc. This study surveys various data classification and data clustering techniques in order to identify the optimal techniques, which provides better outlier predicted data detection. Moreover, the comparison between the various classification and clustering techniques for outlier prediction are illustrated.

並列關鍵字

Data classification ； data clustering ； data mining ； high dimensional data ； outlier detection