Forecasting with Temporal Hierarchies

George Athanasopoulos, Rob J. Hyndman, Nikolaos Kourentzes, Fotios Petropoulos

Research output: Contribution to journalArticle

20 Citations (Scopus)

Abstract

This paper introduces the concept of Temporal Hierarchies for time series forecasting. A temporal hierarchy can be constructed for any time series by means of non-overlapping temporal aggregation. Predictions constructed at all aggregation levels are combined with the proposed framework to result in temporally reconciled, accurate and robust forecasts. The implied combination mitigates modelling uncertainty, while the reconciled nature of the forecasts results in a unified prediction that supports aligned decisions at different planning horizons: from short-term operational up to long-term strategic planning. The proposed methodology is independent of forecasting models. It can embed high level managerial forecasts that incorporate complex and unstructured information with lower level statistical forecasts. Our results show that forecasting with temporal hierarchies increases accuracy over conventional forecasting, particularly under increased modelling uncertainty. We discuss organisational implications of the temporally reconciled forecasts using a case study of Accident & Emergency departments.
LanguageEnglish
Pages60-74
JournalEuropean Journal of Operational Research
Volume262
Issue number1
DOIs
StatusPublished - 1 Oct 2017

Fingerprint

Forecast
Forecasting
Uncertainty Modeling
Temporal Aggregation
Time series
Strategic Planning
Agglomeration
Time Series Forecasting
Prediction
Decision Support
Strategic planning
Accidents
Emergency
Horizon
Aggregation
Planning
Hierarchy
Methodology
Uncertainty modeling
Uncertainty

Cite this

Forecasting with Temporal Hierarchies. / Athanasopoulos, George; Hyndman, Rob J.; Kourentzes, Nikolaos; Petropoulos, Fotios.

In: European Journal of Operational Research, Vol. 262, No. 1, 01.10.2017, p. 60-74.

Research output: Contribution to journalArticle

Athanasopoulos, George ; Hyndman, Rob J. ; Kourentzes, Nikolaos ; Petropoulos, Fotios. / Forecasting with Temporal Hierarchies. In: European Journal of Operational Research. 2017 ; Vol. 262, No. 1. pp. 60-74.
@article{94f16eb5b2e845a589ac30d629c5ede9,
title = "Forecasting with Temporal Hierarchies",
abstract = "This paper introduces the concept of Temporal Hierarchies for time series forecasting. A temporal hierarchy can be constructed for any time series by means of non-overlapping temporal aggregation. Predictions constructed at all aggregation levels are combined with the proposed framework to result in temporally reconciled, accurate and robust forecasts. The implied combination mitigates modelling uncertainty, while the reconciled nature of the forecasts results in a unified prediction that supports aligned decisions at different planning horizons: from short-term operational up to long-term strategic planning. The proposed methodology is independent of forecasting models. It can embed high level managerial forecasts that incorporate complex and unstructured information with lower level statistical forecasts. Our results show that forecasting with temporal hierarchies increases accuracy over conventional forecasting, particularly under increased modelling uncertainty. We discuss organisational implications of the temporally reconciled forecasts using a case study of Accident & Emergency departments.",
author = "George Athanasopoulos and Hyndman, {Rob J.} and Nikolaos Kourentzes and Fotios Petropoulos",
year = "2017",
month = "10",
day = "1",
doi = "10.1016/j.ejor.2017.02.046",
language = "English",
volume = "262",
pages = "60--74",
journal = "European Journal of Operational Research",
issn = "0377-2217",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - Forecasting with Temporal Hierarchies

AU - Athanasopoulos, George

AU - Hyndman, Rob J.

AU - Kourentzes, Nikolaos

AU - Petropoulos, Fotios

PY - 2017/10/1

Y1 - 2017/10/1

N2 - This paper introduces the concept of Temporal Hierarchies for time series forecasting. A temporal hierarchy can be constructed for any time series by means of non-overlapping temporal aggregation. Predictions constructed at all aggregation levels are combined with the proposed framework to result in temporally reconciled, accurate and robust forecasts. The implied combination mitigates modelling uncertainty, while the reconciled nature of the forecasts results in a unified prediction that supports aligned decisions at different planning horizons: from short-term operational up to long-term strategic planning. The proposed methodology is independent of forecasting models. It can embed high level managerial forecasts that incorporate complex and unstructured information with lower level statistical forecasts. Our results show that forecasting with temporal hierarchies increases accuracy over conventional forecasting, particularly under increased modelling uncertainty. We discuss organisational implications of the temporally reconciled forecasts using a case study of Accident & Emergency departments.

AB - This paper introduces the concept of Temporal Hierarchies for time series forecasting. A temporal hierarchy can be constructed for any time series by means of non-overlapping temporal aggregation. Predictions constructed at all aggregation levels are combined with the proposed framework to result in temporally reconciled, accurate and robust forecasts. The implied combination mitigates modelling uncertainty, while the reconciled nature of the forecasts results in a unified prediction that supports aligned decisions at different planning horizons: from short-term operational up to long-term strategic planning. The proposed methodology is independent of forecasting models. It can embed high level managerial forecasts that incorporate complex and unstructured information with lower level statistical forecasts. Our results show that forecasting with temporal hierarchies increases accuracy over conventional forecasting, particularly under increased modelling uncertainty. We discuss organisational implications of the temporally reconciled forecasts using a case study of Accident & Emergency departments.

UR - http://dx.doi.org/10.1016/j.ejor.2017.02.046

U2 - 10.1016/j.ejor.2017.02.046

DO - 10.1016/j.ejor.2017.02.046

M3 - Article

VL - 262

SP - 60

EP - 74

JO - European Journal of Operational Research

T2 - European Journal of Operational Research

JF - European Journal of Operational Research

SN - 0377-2217

IS - 1

ER -