M5 - Mixed reconciliation

Code for reproducing the results of the experiments on the M5 dataset of the paper Probabilistic reconciliation of mixed-type hierarchical time series (Zambon et al., 2024), published at UAI 2024 (the 40th Conference on Uncertainty in Artificial Intelligence). We refer to the paper and to the vignette Reconciliation of M5 hierarchy with mixed-type forecasts of the R package bayesRecon for all the details of the experiments.

Requirements

This is the list of the required packages, available on CRAN:

• bayesRecon
• m5
• parallel
• doSNOW
• foreach
• smooth
• tictoc

Usage

Download the repository and run the file main.R. The folders where data and results are saved can be set in main.R.

Cite Us

@inproceedings{
    zambon2024mixed,
    title={Probabilistic reconciliation of mixed-type hierarchical time series},
    author={Lorenzo Zambon and Dario Azzimonti and Nicolò Rubattu and Giorgio Corani},
    booktitle={The 40th Conference on Uncertainty in Artificial Intelligence},
    year={2024}
}
@Manual{
    bayesRecon,
    title = {bayesRecon: Probabilistic Reconciliation via Conditioning},
    author = {Dario Azzimonti and Nicolò Rubattu and Lorenzo Zambon and Giorgio Corani},
    year = {2023},
    note = {R package version 0.3.0},
  }