New R-package to assess APSIM performance
A new R-package, metrica, has been recently published as part of the CRAN repository (metrica).
This package is an open-source software designed to facilitate the quantitative and visual assessment of prediction performance of point-forecast simulation models for continuous (regression) and categorical variables (classification).
Currently, the assessment of APSIM prediction performance is not integrated for R users, who may rely on a combination of multiple packages and self-defined functions to perform this evaluation.
In this regard, metrica is aimed to simplify this endeavor to APSIM users, allowing to import both APSIM Classic and APSIM Next Generation files, automating the estimation of the most extensive collection of prediction performance metrics (+45) including goodness of fit, error metrics, error decomposition, model efficiencies, indices of agreement, and producing stylish data visualization.
The package includes four example datasets from APSIM (wheat grain N, barley grain number, sorghum grain number, and chickpea dry-mass) that showcase the capabilities of the package for continuous variables.
The full documentation and tutorials are available via GitHub .