Modeling the uncertainty on the covariance matrix for probabilistic forecast reconciliation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Carrara, Chiara, Azzimonti, Dario, Corani, Giorgio, Zambon, Lorenzo
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911659590680576
author Carrara, Chiara
Azzimonti, Dario
Corani, Giorgio
Zambon, Lorenzo
author_facet Carrara, Chiara
Azzimonti, Dario
Corani, Giorgio
Zambon, Lorenzo
contents In minimum trace (MinT) forecast reconciliation, the covariance matrix of the base forecasts errors plays a crucial role. Typically, this matrix is estimated and then treated as known. This can lead to underestimation of the variance of the predictive distribution. To address the problem, we propose a Bayesian reconciliation model that accounts for the uncertainty in the estimation of the covariance matrix. By adopting an Inverse-Wishart prior and assuming Gaussian residuals, the reconciled predictive distribution follows a multivariate t-distribution, obtained in closed-form, rather than a multivariate Gaussian distribution. We evaluate our method on three tourism-related datasets, including a new publicly available dataset. Empirical results show that our approach consistently improves prediction intervals compared to MinT reconciliation.
format Preprint
id arxiv_https___arxiv_org_abs_2506_19554
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Modeling the uncertainty on the covariance matrix for probabilistic forecast reconciliation
Carrara, Chiara
Azzimonti, Dario
Corani, Giorgio
Zambon, Lorenzo
Methodology
Computation
In minimum trace (MinT) forecast reconciliation, the covariance matrix of the base forecasts errors plays a crucial role. Typically, this matrix is estimated and then treated as known. This can lead to underestimation of the variance of the predictive distribution. To address the problem, we propose a Bayesian reconciliation model that accounts for the uncertainty in the estimation of the covariance matrix. By adopting an Inverse-Wishart prior and assuming Gaussian residuals, the reconciled predictive distribution follows a multivariate t-distribution, obtained in closed-form, rather than a multivariate Gaussian distribution. We evaluate our method on three tourism-related datasets, including a new publicly available dataset. Empirical results show that our approach consistently improves prediction intervals compared to MinT reconciliation.
title Modeling the uncertainty on the covariance matrix for probabilistic forecast reconciliation
topic Methodology
Computation
url https://arxiv.org/abs/2506.19554