Correlation-based detection of attribute outliers

An outlier is an object that does not conform to the normal behavior of the data set. In data cleaning, outliers are identified for data noise reduction. In applications such as fraud detection, and stock market analysis, outliers suggest abnormal behavior requiring further investigation. Existing outlier detection methods have focused on class outliers and research on attribute outliers is limited, despite the equal role attribute outliers play in depreciating data quality and reducing data mining accuracy. In this paper, we propose a novel method to detect attribute outliers from the deviating correlation behavior of attributes. We formulate three metrics to evaluate outlier-ness of attributes, and introduce an adaptive factor to distinguish outliers from non-outliers. Experiments with both synthetic and real-world data sets indicate that the proposed method is effective in detecting attribute outliers.