The metrica package compiles over 80 functions for quantifying and visualizing prediction performance of point-forecast models for both continuous (regression) and categorical (classification) targets. Individual metrics such as R2(), RMSE(), and accuracy() share a common obs/pred interface, and metrics_summary() computes a selected list at once.

What’s covered#

  • Basics – obs and pred arguments, data and tidy options shared by all functions
  • Installation – CRAN and GitHub development versions
  • Native datasets – built-in regression (wheat, barley, sorghum, chickpea) and classification (land_cover, maize_phenology) examples
  • Regression metrics – R2(), RMSE(), KGE(), and metrics_summary() for regression
  • Classification metrics – accuracy(), precision(), recall(), fscore(), and metrics_summary() for classification
  • Plots – scatter_plot() and bland_altman_plot()
  • Confusion matrix – confusion_matrix() for binary and multinomial cases