Another look at forecast selection and combination: evidence from forecast pooling

Nikolaos Kourentzes, Devon Barrow, Fotios Petropoulos

Research output: Contribution to journalArticle

Abstract

Forecast selection and combination are regarded as two competing alternatives. In the literature there is substantial evidence that forecast combination is beneficial, in terms of reducing the forecast errors, as well as mitigating modelling uncertainty as we are not forced to choose a single model. However, whether all forecasts to be combined are appropriate, or not, is typically overlooked and various weighting schemes have been proposed to lessen the impact of inappropriate forecasts. We argue that selecting a reasonable pool of forecasts is fundamental in the modelling process and in this context both forecast selection and combination can be seen as two extreme pools of forecasts. We evaluate forecast pooling approaches and find them beneficial in terms of forecast accuracy. We propose a heuristic to automatically identify forecast pools, irrespective of their source or the performance criteria, and demonstrate that in various conditions it performs at least as good as alternative pools that require additional modelling decisions and better than selection or combination.
LanguageEnglish
JournalInternational Journal of Production Economics
Early online date18 May 2018
DOIs
StatusE-pub ahead of print - 18 May 2018

Fingerprint

Pooling
Uncertainty
Forecast combination
Decision modeling
Forecast error
Uncertainty modeling
Performance criteria
Heuristics
Weighting
Forecast accuracy
Process modeling

Cite this

Another look at forecast selection and combination: evidence from forecast pooling. / Kourentzes, Nikolaos; Barrow, Devon; Petropoulos, Fotios.

In: International Journal of Production Economics, 18.05.2018.

Research output: Contribution to journalArticle

@article{aa8d147fab9044a4879c756dbb57ab0b,
title = "Another look at forecast selection and combination: evidence from forecast pooling",
abstract = "Forecast selection and combination are regarded as two competing alternatives. In the literature there is substantial evidence that forecast combination is beneficial, in terms of reducing the forecast errors, as well as mitigating modelling uncertainty as we are not forced to choose a single model. However, whether all forecasts to be combined are appropriate, or not, is typically overlooked and various weighting schemes have been proposed to lessen the impact of inappropriate forecasts. We argue that selecting a reasonable pool of forecasts is fundamental in the modelling process and in this context both forecast selection and combination can be seen as two extreme pools of forecasts. We evaluate forecast pooling approaches and find them beneficial in terms of forecast accuracy. We propose a heuristic to automatically identify forecast pools, irrespective of their source or the performance criteria, and demonstrate that in various conditions it performs at least as good as alternative pools that require additional modelling decisions and better than selection or combination.",
author = "Nikolaos Kourentzes and Devon Barrow and Fotios Petropoulos",
year = "2018",
month = "5",
day = "18",
doi = "10.1016/j.ijpe.2018.05.019",
language = "English",
journal = "International Journal of Production Economics",
issn = "0925-5273",
publisher = "Elsevier",

}

TY - JOUR

T1 - Another look at forecast selection and combination: evidence from forecast pooling

AU - Kourentzes,Nikolaos

AU - Barrow,Devon

AU - Petropoulos,Fotios

PY - 2018/5/18

Y1 - 2018/5/18

N2 - Forecast selection and combination are regarded as two competing alternatives. In the literature there is substantial evidence that forecast combination is beneficial, in terms of reducing the forecast errors, as well as mitigating modelling uncertainty as we are not forced to choose a single model. However, whether all forecasts to be combined are appropriate, or not, is typically overlooked and various weighting schemes have been proposed to lessen the impact of inappropriate forecasts. We argue that selecting a reasonable pool of forecasts is fundamental in the modelling process and in this context both forecast selection and combination can be seen as two extreme pools of forecasts. We evaluate forecast pooling approaches and find them beneficial in terms of forecast accuracy. We propose a heuristic to automatically identify forecast pools, irrespective of their source or the performance criteria, and demonstrate that in various conditions it performs at least as good as alternative pools that require additional modelling decisions and better than selection or combination.

AB - Forecast selection and combination are regarded as two competing alternatives. In the literature there is substantial evidence that forecast combination is beneficial, in terms of reducing the forecast errors, as well as mitigating modelling uncertainty as we are not forced to choose a single model. However, whether all forecasts to be combined are appropriate, or not, is typically overlooked and various weighting schemes have been proposed to lessen the impact of inappropriate forecasts. We argue that selecting a reasonable pool of forecasts is fundamental in the modelling process and in this context both forecast selection and combination can be seen as two extreme pools of forecasts. We evaluate forecast pooling approaches and find them beneficial in terms of forecast accuracy. We propose a heuristic to automatically identify forecast pools, irrespective of their source or the performance criteria, and demonstrate that in various conditions it performs at least as good as alternative pools that require additional modelling decisions and better than selection or combination.

U2 - 10.1016/j.ijpe.2018.05.019

DO - 10.1016/j.ijpe.2018.05.019

M3 - Article

JO - International Journal of Production Economics

T2 - International Journal of Production Economics

JF - International Journal of Production Economics

SN - 0925-5273

ER -