Improving forecasting by estimating time series structural components across multiple frequencies

Nikolaos Kourentzes, Fotios Petropoulos, J.R Trapero Arenas

Research output: Contribution to journalArticle

54 Citations (Scopus)
77 Downloads (Pure)

Abstract

Identifying the most appropriate time series model to achieve a good forecasting accuracy is a challenging task. We propose a novel algorithm that aims to mitigate the importance of model selection, while increasing the accuracy. Multiple time series are constructed from the original time series, using temporal aggregation. These derivative series highlight different aspects of the original data, as temporal aggregation helps in strengthening or attenuating the signals of different time series components. In each series, the appropriate exponential smoothing method is fitted and its respective time series components are forecast. Subsequently, the time series components from each aggregation level are combined, then used to construct the final forecast. This approach achieves a better estimation of the different time series components, through temporal aggregation, and reduces the importance of model selection through forecast combination. An empirical evaluation of the proposed framework demonstrates significant improvements in forecasting accuracy, especially for long-term forecasts.
Original languageEnglish
Pages (from-to)291-302
Number of pages12
JournalInternational Journal of Forecasting
Volume30
Issue number2
Early online date28 Dec 2013
DOIs
Publication statusPublished - 1 Apr 2014

Cite this

Improving forecasting by estimating time series structural components across multiple frequencies. / Kourentzes, Nikolaos; Petropoulos, Fotios; Trapero Arenas, J.R.

In: International Journal of Forecasting, Vol. 30, No. 2, 01.04.2014, p. 291-302.

Research output: Contribution to journalArticle

@article{11f50665e28d4969b82c144d8fa08802,
title = "Improving forecasting by estimating time series structural components across multiple frequencies",
abstract = "Identifying the most appropriate time series model to achieve a good forecasting accuracy is a challenging task. We propose a novel algorithm that aims to mitigate the importance of model selection, while increasing the accuracy. Multiple time series are constructed from the original time series, using temporal aggregation. These derivative series highlight different aspects of the original data, as temporal aggregation helps in strengthening or attenuating the signals of different time series components. In each series, the appropriate exponential smoothing method is fitted and its respective time series components are forecast. Subsequently, the time series components from each aggregation level are combined, then used to construct the final forecast. This approach achieves a better estimation of the different time series components, through temporal aggregation, and reduces the importance of model selection through forecast combination. An empirical evaluation of the proposed framework demonstrates significant improvements in forecasting accuracy, especially for long-term forecasts.",
author = "Nikolaos Kourentzes and Fotios Petropoulos and {Trapero Arenas}, J.R",
year = "2014",
month = "4",
day = "1",
doi = "10.1016/j.ijforecast.2013.09.006",
language = "English",
volume = "30",
pages = "291--302",
journal = "International Journal of Forecasting",
issn = "0169-2070",
publisher = "Elsevier",
number = "2",

}

TY - JOUR

T1 - Improving forecasting by estimating time series structural components across multiple frequencies

AU - Kourentzes, Nikolaos

AU - Petropoulos, Fotios

AU - Trapero Arenas, J.R

PY - 2014/4/1

Y1 - 2014/4/1

N2 - Identifying the most appropriate time series model to achieve a good forecasting accuracy is a challenging task. We propose a novel algorithm that aims to mitigate the importance of model selection, while increasing the accuracy. Multiple time series are constructed from the original time series, using temporal aggregation. These derivative series highlight different aspects of the original data, as temporal aggregation helps in strengthening or attenuating the signals of different time series components. In each series, the appropriate exponential smoothing method is fitted and its respective time series components are forecast. Subsequently, the time series components from each aggregation level are combined, then used to construct the final forecast. This approach achieves a better estimation of the different time series components, through temporal aggregation, and reduces the importance of model selection through forecast combination. An empirical evaluation of the proposed framework demonstrates significant improvements in forecasting accuracy, especially for long-term forecasts.

AB - Identifying the most appropriate time series model to achieve a good forecasting accuracy is a challenging task. We propose a novel algorithm that aims to mitigate the importance of model selection, while increasing the accuracy. Multiple time series are constructed from the original time series, using temporal aggregation. These derivative series highlight different aspects of the original data, as temporal aggregation helps in strengthening or attenuating the signals of different time series components. In each series, the appropriate exponential smoothing method is fitted and its respective time series components are forecast. Subsequently, the time series components from each aggregation level are combined, then used to construct the final forecast. This approach achieves a better estimation of the different time series components, through temporal aggregation, and reduces the importance of model selection through forecast combination. An empirical evaluation of the proposed framework demonstrates significant improvements in forecasting accuracy, especially for long-term forecasts.

U2 - 10.1016/j.ijforecast.2013.09.006

DO - 10.1016/j.ijforecast.2013.09.006

M3 - Article

VL - 30

SP - 291

EP - 302

JO - International Journal of Forecasting

JF - International Journal of Forecasting

SN - 0169-2070

IS - 2

ER -