Home > Research > Publications & Outputs > Neural network ensemble operators for time seri...

Electronic data

Links

Text available via DOI:

View graph of relations

Neural network ensemble operators for time series forecasting

Research output: Contribution to journalJournal articlepeer-review

Published

Standard

Neural network ensemble operators for time series forecasting. / Kourentzes, Nikos; Barrow, Devon; Crone, Sven.

In: Expert Systems with Applications, Vol. 41, No. 9, 07.2014, p. 4235-4244.

Research output: Contribution to journalJournal articlepeer-review

Harvard

APA

Vancouver

Author

Kourentzes, Nikos ; Barrow, Devon ; Crone, Sven. / Neural network ensemble operators for time series forecasting. In: Expert Systems with Applications. 2014 ; Vol. 41, No. 9. pp. 4235-4244.

Bibtex

@article{df8f3e0f0729449fbd6ff8085b9be175,
title = "Neural network ensemble operators for time series forecasting",
abstract = "The combination of forecasts resulting from an ensemble of neural networks has been shown to outperform the use of a single ``best'' network model. This is supported by an extensive body of literature, which shows that combining generally leads to improvements in forecasting accuracy and robustness, and that using the mean operator often outperforms more complex methods of combining forecasts. This paper proposes a mode ensemble operator based on kernel density estimation, which unlike the mean operator is insensitive to outliers and deviations from normality, and unlike the median operator does not require symmetric distributions. The three operators are compared empirically and the proposed mode ensemble operator is found to produce the most accurate forecasts, followed by the median, while the mean has relatively poor performance. The findings suggest that the mode operator should be considered as an alternative to the mean and median operators in forecasting applications. Experiments indicate that mode ensembles are useful in automating neural network models across a large number of time series, overcoming issues of uncertainty associated with data sampling, the stochasticity of neural network training and the distribution of the forecasts.",
keywords = "Time Series, Forecasting , Ensembles , Combination , Mode Estimation , Kernel Density Estimation , Neural Networks , Mean , Median",
author = "Nikos Kourentzes and Devon Barrow and Sven Crone",
note = "The final, definitive version of this article has been published in the Journal, Expert Systems with Applications 41 (9), 2014, {\textcopyright} ELSEVIER.",
year = "2014",
month = jul,
doi = "10.1016/j.eswa.2013.12.011",
language = "English",
volume = "41",
pages = "4235--4244",
journal = "Expert Systems with Applications",
issn = "0957-4174",
publisher = "Elsevier Ltd",
number = "9",

}

RIS

TY - JOUR

T1 - Neural network ensemble operators for time series forecasting

AU - Kourentzes, Nikos

AU - Barrow, Devon

AU - Crone, Sven

N1 - The final, definitive version of this article has been published in the Journal, Expert Systems with Applications 41 (9), 2014, © ELSEVIER.

PY - 2014/7

Y1 - 2014/7

N2 - The combination of forecasts resulting from an ensemble of neural networks has been shown to outperform the use of a single ``best'' network model. This is supported by an extensive body of literature, which shows that combining generally leads to improvements in forecasting accuracy and robustness, and that using the mean operator often outperforms more complex methods of combining forecasts. This paper proposes a mode ensemble operator based on kernel density estimation, which unlike the mean operator is insensitive to outliers and deviations from normality, and unlike the median operator does not require symmetric distributions. The three operators are compared empirically and the proposed mode ensemble operator is found to produce the most accurate forecasts, followed by the median, while the mean has relatively poor performance. The findings suggest that the mode operator should be considered as an alternative to the mean and median operators in forecasting applications. Experiments indicate that mode ensembles are useful in automating neural network models across a large number of time series, overcoming issues of uncertainty associated with data sampling, the stochasticity of neural network training and the distribution of the forecasts.

AB - The combination of forecasts resulting from an ensemble of neural networks has been shown to outperform the use of a single ``best'' network model. This is supported by an extensive body of literature, which shows that combining generally leads to improvements in forecasting accuracy and robustness, and that using the mean operator often outperforms more complex methods of combining forecasts. This paper proposes a mode ensemble operator based on kernel density estimation, which unlike the mean operator is insensitive to outliers and deviations from normality, and unlike the median operator does not require symmetric distributions. The three operators are compared empirically and the proposed mode ensemble operator is found to produce the most accurate forecasts, followed by the median, while the mean has relatively poor performance. The findings suggest that the mode operator should be considered as an alternative to the mean and median operators in forecasting applications. Experiments indicate that mode ensembles are useful in automating neural network models across a large number of time series, overcoming issues of uncertainty associated with data sampling, the stochasticity of neural network training and the distribution of the forecasts.

KW - Time Series

KW - Forecasting

KW - Ensembles

KW - Combination

KW - Mode Estimation

KW - Kernel Density Estimation

KW - Neural Networks

KW - Mean

KW - Median

U2 - 10.1016/j.eswa.2013.12.011

DO - 10.1016/j.eswa.2013.12.011

M3 - Journal article

VL - 41

SP - 4235

EP - 4244

JO - Expert Systems with Applications

JF - Expert Systems with Applications

SN - 0957-4174

IS - 9

ER -