Diagnostic Graphics to Evaluate Forecast Performance
Accessing original forecast and observation data for triptych objects
Adding confidence regions
Adding consistency regions for reliability curves
Accessing diagnostic estimate data
Evaluation of forecasts using score decompositions
Evaluation of forecasts using Murphy curves
Plot methods for the triptych classes
Objects exported from other packages
Accessing confidence/consistency region data
Evaluation of forecasts using reliability curves
Bootstrap (binary) observation resampling for triptych objects
Bootstrap case resampling for triptych objects
Evaluation of forecasts using ROC curves
Internal vctrs methods
Evaluation of forecasts using a Triptych
Overall predictive performance is measured by a mean score (or loss), which decomposes into miscalibration, discrimination, and uncertainty components. The main focus is visualization of these distinct and complementary aspects in joint displays. See Dimitriadis, Gneiting, Jordan, Vogel (2024) <doi:10.1016/j.ijforecast.2023.09.007>.