fable.bayesRecon: BAyesian reCONciliation in the fable framework

The package fable.bayesRecon integrates the probabilistic reconciliation methods from bayesRecon into the fable / fabletools framework. Reconciliation is specified via the reconcile() verb and produced when forecast() is called, following the same tidy workflow used by fable.

The reconciliation functions are:

• bayesRecon_t: reconciliation via conditioning with uncertain covariance matrix; the reconciled forecasts are multivariate Student-t; this is done analytically.
• bayesRecon_BUIS: reconciliation via conditioning of any probabilistic forecast via importance sampling; this is the recommended option for non-Gaussian base forecasts;
• bayesRecon_MixCond: reconciliation via conditioning of mixed hierarchies, where the upper forecasts are multivariate Gaussian and the bottom forecasts are discrete distributions;
• bayesRecon_TDcond: reconciliation via top-down conditioning of mixed hierarchies, where the upper forecasts are multivariate Gaussian and the bottom forecasts are discrete distributions;

News

💥 [2026-05-05] fable.bayesRecon v0.1.0: first CRAN release.

Installation

You can install the development version from GitHub:

# install.packages("devtools")
devtools::install_github("dazzimonti/fable.bayesRecon", build_vignettes = TRUE, dependencies = TRUE)

Usage

The package follows the standard fable workflow:

1. Prepare data as a tsibble and define the hierarchy with aggregate_key().
2. Fit base forecasting models with model().
3. Specify the reconciliation strategy inside reconcile().
4. Produce reconciled probabilistic forecasts with forecast().

We provide in this vignette a simple usage example; refer to the package documentation for more details on the reconciliation methods and their parameters. See the book Hyndman and Athanasopoulos (2021) for a general introduction to forecasting with fable and fabletools.

References

Carrara, C., Corani, G., Azzimonti, D., Zambon, L. (2025). Modeling the uncertainty on the covariance matrix for probabilistic forecast reconciliation. arXiv preprint arXiv:2506.19554. Available here

Hyndman, R.J., & Athanasopoulos, G. (2021). Forecasting: principles and practice. 3rd edition, OTexts: Melbourne, Australia. OTexts.com/fpp3. Accessed on 05/05/2026.

Zambon, L., Azzimonti, D. & Corani, G. (2024). Efficient probabilistic reconciliation of forecasts for real-valued and count time series. Statistics and Computing 34 (1), 21. DOI

Zambon, L., Azzimonti, D., Rubattu, N., Corani, G. (2024). Probabilistic reconciliation of mixed-type hierarchical time series. Proceedings of the Fortieth Conference on Uncertainty in Artificial Intelligence, PMLR 244:4078-4095. Available here

Contributors

Dario Azzimonti (Maintainer) Email

Stefano Damato   Email

Lorenzo Zambon   Email

Chiara Carrara   Email

Giorgio Corani   Email

Getting help

If you encounter a bug, please file a minimal reproducible example on GitHub.