Querying and mining data streams: you only get one look a tutorial
Proceedings of the 2002 ACM SIGMOD international conference on Management of data
A Unifying Framework for Detecting Outliers and Change Points from Time Series
IEEE Transactions on Knowledge and Data Engineering
Unsupervised Outlier Detection in Time Series Data
ICDEW '06 Proceedings of the 22nd International Conference on Data Engineering Workshops
Hi-index | 0.00 |
Online detection of outliers and change points from a data stream are two very exciting topics in the area of data mining. This paper explores the relationship between these two issues, and presents a unifying method for dealing with both of them. Previous approaches often use parametric techniques and try to give exact results. In contrast, we present a nonparametric method based on local polynomial fitting, and give approximate results by fuzzy partition and decision. In order to measure the possibility of being an outlier and a change point, two novel score functions are defined based on the forward and backward prediction errors. The proposed method can detect outliers and changes simultaneously, and can distinguish between them. Comparing to the conventional parametric approaches, our method is more convenient for implementation, and more appropriate for online and interactive data mining. Simulation results confirm the effectiveness of the proposed method.