This function calculates the RegressionROC Curve of of Hernández-Orallo doi:10.1016/j.patcog.2013.06.014 . It provides estimates for the positive and negative errors when predictions are shifted by a variety of constants (which range across the domain of observed residuals). Curves closer to the axes are, in general, to be preferred. In general, this curve provides a simple way to visualize the error properties of a regression model.
calculate_rroc(label, prediction, nbins = 100)True label
Model prediction of the label (out of sample)
Number of shift values to sweep over
A tibble with nbins rows.
The dot shows the errors when no shift is applied, corresponding to the base model predictions.
Hernández-Orallo, J. (2013). ROC curves for regression. Pattern Recognition, 46(12), 3395-3411.