An efficient approach for mining segment-wise intervention rules in time-series streams

  • Authors:
  • Yue Wang;Jie Zuo;Ning Yang;Lei Duan;Hong-Jun Li;Jun Zhu

  • Affiliations:
  • DB&KE Lab., School of Computer Science, Sichuan University, Chengdu, China;DB&KE Lab., School of Computer Science, Sichuan University, Chengdu, China;DB&KE Lab., School of Computer Science, Sichuan University, Chengdu, China;DB&KE Lab., School of Computer Science, Sichuan University, Chengdu, China;DB&KE Lab., School of Computer Science, Sichuan University, Chengdu, China;China Birth Defect Monitoring Centre, Sichuan University, Chengdu, China

  • Venue:
  • WAIM'10 Proceedings of the 11th international conference on Web-age information management
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Huge time-series stream data are collected every day from many areas, and their trends may be impacted by outside events, hence biased from its normal behavior. This phenomenon is referred as intervention. Intervention rule mining is a new research direction in data mining with great challenges. To solve these challenges, this study makes the following contributions: (a) Proposes a framework to detect intervention events in time-series streams, (b) Proposes approaches to evaluate the impact of intervention events, and (c) Conducts extensive experiments both on real data and on synthetic data. The results of the experiments show that the newly proposed methods reveal interesting knowledge and perform well with good accuracy and efficiency.