Skip to content

JulienLeprince/hierarchicallearning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

hierarchicallearning

Hierarchical Learning

This repository presents Hierarchical Learning, a package for hierarchical forecasting of time series developed in the open-access journal article Hierarchical learning, forecasting coherent spatio-temporal individual and aggregated building loads. The repository reports both the python implementations of the article using the open data set of the building-data-genome-project-2 as well as smaller tutorial jupiter notebooks to illustrate usage of the developed classes.

To get started, simply clone this repository on your computer or Fork it via GitHub. After installing dependencies from the requirements.txt file, the code should run properly.

Citation

If you find this code useful and use it in your work, please reference our journal article:

Leprince, J., Madsen, H., Kloppenborg Møller, J. and Zeiler, W., 2023. Hierarchical learning, forecasting coherent spatio-temporal individual and aggregated building loads. Applied Energy, p.112095.

@article{LEPRINCE2023121510,
title = {Hierarchical learning, forecasting coherent spatio-temporal individual and aggregated building loads},
journal = {Applied Energy},
volume = {348},
pages = {121510},
year = {2023},
issn = {0306-2619},
doi = {https://doi.org/10.1016/j.apenergy.2023.121510},
url = {https://www.sciencedirect.com/science/article/pii/S0306261923008747},
author = {Julien Leprince and Henrik Madsen and Jan Kloppenborg Møller and Wim Zeiler},
}

Repository structure

hierarchical learning
├─ fig                                              <- repository readme figures
├─ io
|   ├─ input                                        <- input data-set of the Fox site
|   └─ output
|       ├─ regressor                                <- fitted hierarchical models and saved performances
|       ├─ results                                  <- processed results for forecast performance plots
|       └─ visuals                                  <- forecast performance plots
├─ src
|   ├─ hl                                           <- hierarchical learning package
|   ├─ publication
|   |   ├─ 0 preprocessing                          <- BDG2 data preprocessing
|   |   ├─ 1 cluster2tree                           <- reducing spatial hierarchical dimension with clustering
|   |   ├─ 2 hierarchical forecasting <dimension>   <- hierarchical learning and reconciliation
|   |   ├─ 3 base forecasting <dimension>           <- base forecast and reconciliation
|   |   ├─ 4 results processing                     <- forecast performance result processing
|   |   └─ 5 results visualization                  <- forecast performance visualizations
|   └─ tutorial
|       ├─ 0 manipulating trees                     <- tree class initialization exemplified on all dimensions
|       └─ 1 forecasting trees                      <- tree regressor initialization and training
└─ README.md                                        <- README for developers using this code

Authors

Julien Leprince, Prof. Henrik Madsen, Ass. Prof. Jan Kloppenborg Møller, Prof. Wim Zeiler.

License

This project is licensed under the MIT License - see the LICENSE.md file for details

About

Hierarchical forecasting with machine learning

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published