Advertisement

Moments of Predictive Deviations for Ensemble Diversity Measures to Estimate the Performance of Time Series Prediction

  • Kohei Ono
  • Shuichi Kurogi
  • Takeshi Nishida
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 7667)

Abstract

This paper presents an analysis of moments of predictive deviations as measures of ensemble diversity to estimate the performance of time series prediction. As an extension of the ambiguity decomposition of bagging ensemble, we decompose the fourth power of ensemble prediction error and clarify the effect of the moments of predictive deviations of ensemble members to the ensemble prediction error. We utilize this analysis for estimating the performance of time sires prediction, and show the effectiveness by means of numerical experiments.

Keywords

Moments of predictive deviations Ensemble diversity measures Performance estimation Time series prediction 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Brown, G., Wyatt, J., Tino, P.: Managing Diversity in Regression Ensembles. J. Mach. Learn. Res. 6, 1621–1650 (2005)MathSciNetzbMATHGoogle Scholar
  2. 2.
    Chen, H.: Diversity and Regularization in Neural Network Ensembles. PHD thesis, University of Birmingham (2008)Google Scholar
  3. 3.
    Kurogi, S.: Improving Generalization Performance via out-of-Bag Estimate Using Variable Size of Bags. J. Japan. Neural Network Society 16, 81–92 (2009)CrossRefGoogle Scholar
  4. 4.
    Kohavi, R.: A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection. In: Proceedings of the Fourteenth International Conference 18 on Artificial Intelligence (IJCAI), pp. 1137–1143 (1995)Google Scholar
  5. 5.
    Efron, B., Tbshirani, R.: Improvements on Cross-Validation: the.632+ bootstrap method. J. American Stats. Associ. 92, 548–560 (1997)zbMATHGoogle Scholar
  6. 6.
    Breiman, L.: Bagging Predictors. Mach. Learn. 26, 123–140 (1996)Google Scholar
  7. 7.
    Aihara, K.: Theories and Applications of Chaotic Time Series Analysis, Sangyo Tosho, Tokyo (2000)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  • Kohei Ono
    • 1
  • Shuichi Kurogi
    • 1
  • Takeshi Nishida
    • 1
  1. 1.Kyushu Institute of technologyFukuokaJapan

Personalised recommendations