Dimensionality reduction in forecasting with temporal hierarchies
Nystrup, Peter LU
; Lindström, Erik
LU
; Møller, Jan K.
and Madsen, Henrik
(2021)
In International Journal of Forecasting
37(3).
p.1127-1146
- Abstract
Combining forecasts from multiple temporal aggregation levels exploits information differences and mitigates model uncertainty, while reconciliation ensures a unified prediction that supports aligned decisions at different horizons. It can be challenging to estimate the full cross-covariance matrix for a temporal hierarchy, which can easily be of very large dimension, yet it is difficult to know a priori which part of the error structure is most important. To address these issues, we propose to use eigendecomposition for dimensionality reduction when reconciling forecasts to extract as much information as possible from the error structure given the data available. We evaluate the proposed estimator in a simulation study and demonstrate... (More)
Combining forecasts from multiple temporal aggregation levels exploits information differences and mitigates model uncertainty, while reconciliation ensures a unified prediction that supports aligned decisions at different horizons. It can be challenging to estimate the full cross-covariance matrix for a temporal hierarchy, which can easily be of very large dimension, yet it is difficult to know a priori which part of the error structure is most important. To address these issues, we propose to use eigendecomposition for dimensionality reduction when reconciling forecasts to extract as much information as possible from the error structure given the data available. We evaluate the proposed estimator in a simulation study and demonstrate its usefulness through applications to short-term electricity load and financial volatility forecasting. We find that accuracy can be improved uniformly across all aggregation levels, as the estimator achieves state-of-the-art accuracy while being applicable to hierarchies of all sizes.
(Less)
- author
- Nystrup, Peter
LU
; Lindström, Erik
LU
; Møller, Jan K.
and Madsen, Henrik
- organization
- publishing date
- 2021-01-10
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Load forecasting, Realized volatility, Reconciliation, Shrinkage, Spectral decomposition, Temporal aggregation
- in
- International Journal of Forecasting
- volume
- 37
- issue
- 3
- pages
- 20 pages
- publisher
- Elsevier
- external identifiers
-
- scopus:85099146966
- ISSN
- 0169-2070
- DOI
- 10.1016/j.ijforecast.2020.12.003
- language
- English
- LU publication?
- yes
- id
- 5c60fa83-3125-4752-b9a7-efe4c7d52bac
- date added to LUP
- 2021-01-20 10:56:21
- date last changed
- 2025-04-04 14:54:33
@article{5c60fa83-3125-4752-b9a7-efe4c7d52bac,
abstract = {{<p>Combining forecasts from multiple temporal aggregation levels exploits information differences and mitigates model uncertainty, while reconciliation ensures a unified prediction that supports aligned decisions at different horizons. It can be challenging to estimate the full cross-covariance matrix for a temporal hierarchy, which can easily be of very large dimension, yet it is difficult to know a priori which part of the error structure is most important. To address these issues, we propose to use eigendecomposition for dimensionality reduction when reconciling forecasts to extract as much information as possible from the error structure given the data available. We evaluate the proposed estimator in a simulation study and demonstrate its usefulness through applications to short-term electricity load and financial volatility forecasting. We find that accuracy can be improved uniformly across all aggregation levels, as the estimator achieves state-of-the-art accuracy while being applicable to hierarchies of all sizes.</p>}},
author = {{Nystrup, Peter and Lindström, Erik and Møller, Jan K. and Madsen, Henrik}},
issn = {{0169-2070}},
keywords = {{Load forecasting; Realized volatility; Reconciliation; Shrinkage; Spectral decomposition; Temporal aggregation}},
language = {{eng}},
month = {{01}},
number = {{3}},
pages = {{1127--1146}},
publisher = {{Elsevier}},
series = {{International Journal of Forecasting}},
title = {{Dimensionality reduction in forecasting with temporal hierarchies}},
url = {{http://dx.doi.org/10.1016/j.ijforecast.2020.12.003}},
doi = {{10.1016/j.ijforecast.2020.12.003}},
volume = {{37}},
year = {{2021}},
}