Intervention Events Detection and Prediction in Data Streams

Abstract. Mining interesting patterns in data streams has attracted special attention recently. This study revealed the principles behind observations, through variation of intervention events to analyze the trends in the data streams. The main contributions includes: (a) Proposed a novel concept intervention event, and method to analyze streams under intervention. (b) Proposed the methods to evaluate the impact of intervention events. (c) Gave extensive experiments on real data to show that the newly proposed methods do prediction efficiently, and the rate of success is almost reach 92.6% recall in adaptive detection for intervention events in practical environment.

• Type: Conference
• Year: 2009
• Where: WAIM
• Authors: Yue Wang, Changjie Tang, Chuan Li, Yu Chen, Ning Yang, Rong Tang, Jun Zhu
• Keywords: Data Streams, Intervention Events, Mining Interesting Patterns, WAIM 2009