Outlier Detection Based On EP Pattern

Research outliers research in many fields, such as finance and information gathering have been widespread concern. People often think outlier data has changed the original data so that the data set or change the data generating mechanism. Therefore, the excavation of some outliers become a significant work, especially in the actual operation of stock trading, credit card fraud transactions, especially in the obvious. Outlier mining goal is to find those with the dataset most objects have very different behavior of rare data, research outliers abnormal behavior can help uncover hidden valuable knowledge, assist decision makers, gain or improve the quality of service.Firstly, from all aspects of practical significance outliers, algorithms, applications and other issues to find outliers were reviewed. Analysis of the real problems in the data, summarized data processing methods for different problems adopted.Secondly, the paper introduces the EP mode, summed up the reasonableness and necessity of EP mode algorithms exist; to study the EP mode algorithm based on outlier high-dimensional data mining algorithms are described. EP mode algorithms and other high-dimensional data processing algorithms were compared against the distance scale and regional density between high-dimensional data no longer has an intuitive sense and correct rate of decline in mining algorithms and other issues, this paper based on the EP mode of outliers discovery algorithm to achieve high dimensional data quickly find outliers.The paper outlier algorithm based on EP mode has been fully explained, elaborated algorithm based on outlier EP mode advantages and characteristics. Algorithm advantage in high-dimensional data. Finally, the algorithm is implemented based on outlier EP mode EP mode based outlier System.