FoRecoML

Forecast Reconciliation is a post-forecasting process designed to improve accuracy and align forecasts within systems of linearly constrained time series (e.g. hierarchical or grouped). The FoRecoML package provides nonlinear forecast reconciliation procedures using Machine Learning in cross-sectional, temporal, and cross-temporal settings. FoRecoML inherits time series processing functionalities from FoReco.

The core functions for reconciliation are:

• csrml() Cross-sectional Reconciliation with Machine Learning

• terml() Temporal Reconciliation with Machine Learning

• ctrml() Cross-temporal Reconciliation with Machine Learning

• extract_reconciled_ml() Extraction of the fitted machine learning model used for forecast reconciliation from the output of one of the reconciliation function. The fitted machine learning model can be reused for different sets of data with the same hierarchical structure.

Machine learning models that can be used with FoRecoML include random forest (randomForest), extreme gradient boosting (xgboost), light gradient boosting machine (lightgbm), and models supported by the mlr3 package.

Installation

You can install the stable version on CRAN

install.packages("FoRecoML")

You can install the development version of FoRecoML from GitHub

# install.packages("devtools")
devtools::install_github("danigiro/FoRecoML")

Code of Conduct

Please note that the FoRecoML project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.