An outlier mining algorithm based on confidence interval

Outlier detection is a hot topic of data mining. After studying the existing classical algorithms of detecting outlier, this paper proposes an outlier mining algorithm based on confidence interval, and makes a new definition for outlier. The method combines mathematical statistics and density-based clustering algorithm. It clustering firstly with DBSCAN algorithm, obtains credible sample and suspicious outliers. Secondly, a confidence interval is obtained based on credible sample, then suspicious outliers will be detected and disposed using the confidence interval. The experiment results on IRIS show that this algorithm can detect outliers effectively.