Journal of Frontiers of Computer Science and Technology ›› 2014, Vol. 8 ›› Issue (10): 1153-1161.DOI: 10.3778/j.issn.1673-9418.1407035

Previous Articles     Next Articles

SPG-Suite: Forecasting Method Towards Pseudo Periodic Time Series

HONG Shenda1,2, YIN Ning1,2, QIU Zhen1,2, FAN Lilue1,2, LI Hongyan1,2+   

  1. 1. School of Electronics Engineering and Computer Science, Peking University, Beijing 100871, China
    2. Key Laboratory of Machine Perception (Ministry of Education), Peking University, Beijing 100871, China
  • Online:2014-10-01 Published:2014-09-29

SPG-Suite:面向伪周期时间序列的预测方法

洪申达1,2,尹  宁1,2,邱  镇1,2,樊里略1,2,李红燕1,2+   

  1. 1. 北京大学 信息科学技术学院,北京 100871
    2. 北京大学 机器感知与智能教育部重点实验室,北京 100871

Abstract: Pseudo periodic time series is a kind of widespread data form with features including pseudo periodicity, non-stationary and transition values. Forecasting this kind of data is significantly important on both research and application fields. However, current research pays insufficient attention on pseudo periodic time series. Forecasting errors are cumulated when using the existing methods in pseudo periodic time series, which makes the prediction effect poor. This paper firstly analyzes and summarizes the features of pseudo periodic time series. Then, it proposes SPG-Suite forecasting method to solve the specific pseudo periodic time series problem. The experimental results on the real datasets show that the SPG-Suite method has better performance on forecasting accuracy and scalability compared with the existing methods.

Key words: time series, pseudo period, forecasting

摘要: 伪周期时间序列是一种广泛存在的数据形式,它具有伪周期性、非平稳性和特征值等特征。对这类时间序列进行预测,具有很强的研究和应用意义。然而,目前的相关研究对伪周期时间序列的关注度不足,一些已有的时间序列预测方法在应用到伪周期时间序列上时,会造成误差的累积,使得预测效果很差。为了解决这些问题,总结了伪周期时间序列的特征,并提出了SPG-Suite预测方法,很好地解决了传统方法无法解决的问题。最后,在真实的数据集上进行了实验,并与传统方法进行了对比,实验结果表明,SPG-Suite方法在预测精度上具有明显的优势,并具有较强的可扩展性。

关键词: 时间序列, 伪周期, 预测