Numeric metrics evaluate regression predictions where both truth and
estimate are numeric. These metrics measure how close predicted values
are to the true values.
Available metrics
ccc()Direction: maximize. Range: [-1, 1]
huber_loss()Direction: minimize. Range: [0, Inf]
huber_loss_pseudo()Direction: minimize. Range: [0, Inf]
iic()Direction: maximize. Range: [-1, 1]
mae()Direction: minimize. Range: [0, Inf]
mape()Direction: minimize. Range: [0, Inf]
mase()Direction: minimize. Range: [0, Inf]
mpe()Direction: zero. Range: [-Inf, Inf]
msd()Direction: zero. Range: [-Inf, Inf]
poisson_log_loss()Direction: minimize. Range: [0, Inf]
rmse()Direction: minimize. Range: [0, Inf]
rpd()Direction: maximize. Range: [0, Inf]
rpiq()Direction: maximize. Range: [0, Inf]
rsq()Direction: maximize. Range: [-Inf, 1]
rsq_trad()Direction: maximize. Range: [0, 1]
smape()Direction: minimize. Range: [0, 100]
See also
quantile-metrics for quantile prediction metrics
vignette("metric-types") for an overview of all metric types
Examples
data("solubility_test")
head(solubility_test)
#> solubility prediction
#> 1 0.93 0.3677522
#> 2 0.85 -0.1503220
#> 3 0.81 -0.5051844
#> 4 0.74 0.5398116
#> 5 0.61 -0.4792718
#> 6 0.58 0.7377222
rmse(solubility_test, solubility, prediction)
#> # A tibble: 1 × 3
#> .metric .estimator .estimate
#> <chr> <chr> <dbl>
#> 1 rmse standard 0.722