Applying Computational Intelligence Techniques to QoS Time Series Forecasting in Services Computing - Volume 13 Number 9 (Sep. 2018) - JCOMPUTERS
Volume 13 Number 9 (Sep. 2018)
Home > Archive > 2018 > Volume 13 Number 9 (Sep. 2018) >
JCP 2018 Vol.13(9): 1010-1026 ISSN: 1796-203X
doi: 10.17706/jcp.13.9.1010-1026

Applying Computational Intelligence Techniques to QoS Time Series Forecasting in Services Computing

Yang Syu1, Yong-Yi FanJiang2
1Institute of Information Science, Academia Sinica, Taipei City, Taiwan.
2Department of Computer Science and Information Engineering, Fu Jen Catholic University, New Taipei City, Taiwan


Abstract—Recently, the forecasting of dynamic Quality of Service (QoS) values for Web Services (WSs) has become an emerging topic in services computing. In most previous research, various time series forecasting methods have been used to address this problem. In this paper, we propose the use of two computational intelligence techniques, namely, genetic programming (GP) and support vector regression (SVR). To demonstrate the forecasting performance of the two proposed techniques, we compare them with the conventional methods based on experiments run on a real-world dynamic QoS time series dataset. Our experimental results show that the proposed GP and SVR methods outperform the conventional methods in both training (in-sample) and testing (out-of-sample) accuracy. Between the two proposed approaches, we find that GP might be the better choice overall. In terms of training performance, GP is superior in terms of both the individual and average experimental results; however, this is not the case for testing performance. In terms of testing accuracy, SVR outperforms GP in many individual experiments; however, SVR also yields extremely poor forecasting accuracy in several individual experiments, indicating that it is unstable and unreliable. In many of the individual experiments, GP is only insignificantly inferior to SVR, and it still achieves the best average forecasting accuracy according to two of the three considered measures.

Index Terms—Time series forecasting, quality of service, computational intelligence techniques, support vector regression, genetic programming.

[PDF]

Cite: Yang Syu, Yong-Yi FanJiang, "Applying Computational Intelligence Techniques to QoS Time Series Forecasting in Services Computing," Journal of Computers vol. 13, no. 9, pp. 1010-1026, 2018.

General Information

ISSN: 1796-203X
Frequency: Monthly (2006-2014); Bimonthly (Since 2015)
Editor-in-Chief: Prof. Liansheng Tan
Executive Editor: Ms. Nina Lee
Abstracting/ Indexing: DBLP, EBSCO,  ProQuest, INSPEC, ULRICH's Periodicals Directory, WorldCat, CNKI,etc
E-mail: jcp@iap.org
  • Aug 06, 2018 News!

    Vol 13, No 1-No 8 has been indexed by EI (Inspec)   [Click]

  • Aug 06, 2018 News!

    Vol 12, No 6 has been indexed by EI (Inspec)   [Click]

  • Apr 24, 2018 News!

    Vol 13, No 9 has been published with online version   [Click]

  • Dec 26, 2017 News!

    Vol 12, No 1-No 5 has been indexed by EI (Inspec)     [Click]

  • Dec 26, 2017 News!

    Vol 11, No 4-No 6 has been indexed by EI (Inspec)     [Click]

  • Read more>>