Outlier detection scoring measurements based on frequent pattern technique

Outlier detection is one of the main data mining tasks. The outliers in data are more significant and interesting than common ones in a wide variety of application domains, such as fraud detection, intrusion detection, ecosystem disturbances and many others. Recently, a new trend for detecting the outlier by discovering frequent patterns (or frequent item sets) from the data set has been studied. In this study, we present a summarization and comparative study of the available outlier detection scoring measurements which are based on the frequent patterns discovery. The comparisons of the outlier detection scoring measurements are based on the detection effectiveness. The results of the comparison prove that this approach of outlier detection is a promising approach to be utilized in different domain applications.