Relative Mean Difference (RMD)

Computes the relative mean signed deviation between `estimate` and `truth`, calculated as the mean of (`estimate - truth`) divided by the mean of `truth`, expressed as a percentage. Positive values indicate overestimation.

This is a thin wrapper around `yardstick::msd()`, with reversed sign and scaled by the mean of the true values. The default `yardstick::msd()` computes `truth - estimate`, while this function computes `estimate - truth`.

Usage

rmd(data, ...)

# S3 method for class 'data.frame'
rmd(data, truth, estimate, na_rm = TRUE, case_weights = NULL, ...)

rmd_vec(truth, estimate, na_rm = TRUE, case_weights = NULL, ...)

Arguments

data: A data frame containing the `truth` and `estimate` columns.
...: Additional arguments passed to lower-level functions.
truth: The column identifier for the true values (bare or quoted).
estimate: The column identifier for the predicted values (bare or quoted).
na_rm: A logical value indicating whether `NA` values should be removed before computation.
case_weights: An optional column of case weights.

Details

The relative mean difference is useful when comparing average deviations across models or datasets, as it expresses the difference relative to the size of the true value. It supports optional case weights and handles missing values similarly to other `yardstick` metrics.