Compute Agreement Metrics for Predictions — agreement

This function calculates various metrics to assess the agreement between predicted and true values. It leverages the flexibility of the `yardstick` package to compute a variety of metrics, which can be specified through the `metrics` parameter.

Usage

agreement_metrics(
  data,
  truth,
  estimate,
  metrics = list(rsq, md, rmd, rmse, rrmse),
  label = FALSE
)

Arguments

data: A data frame containing the observed (`truth`) and predicted (`estimate`) values.
truth: The column in `data` representing the true values. Use tidy evaluation (e.g., ``).
estimate: The column in `data` representing the predicted values. Use tidy evaluation (e.g., ``).
metrics: A list of metrics to compute. Metrics can be almost any function from the `yardstick` package (e.g., `rsq`, `rmse`, `mape`, `msd`), provided they follow the `yardstick` format. Defaults to `list(rsq, md, rmd, rmse, rrmse)`. Can also be a named list of metrics to compute (e.g. `list(bias=md, "bias Names are then used instead of the `yardstick` metric names.
label: Logical. If `TRUE`, the function creates a concatenated string summarizing all computed metrics in a new column called `label`. Defaults to `FALSE`.

Value

A data frame with the computed metrics. If `label = TRUE`, the output includes a `label` column summarizing all metrics.

Details

The function internally uses the `metric_set` function from `yardstick` to handle the specified metrics. The results are returned in a wide format, with each metric as a column. All numeric metrics are rounded to two decimal places for readability.

Examples

library(dplyr)
#> 
#> Attaching package: ‘dplyr’
#> The following objects are masked from ‘package:stats’:
#> 
#>     filter, lag
#> The following objects are masked from ‘package:base’:
#> 
#>     intersect, setdiff, setequal, union
library(yardstick)
library(tidyr)

# Example dataset
data <- data.frame(
  truth = c(3, 5, 2.5, 7),
  estimate = c(2.8, 5.1, 2.6, 7.2)
)

# Compute default metrics
agreement_metrics(data, truth = truth, estimate = estimate)
#> # A tibble: 1 × 5
#>     rsq    md   rmd  rmse rrmse
#>   <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1     1  0.05  1.14  0.16  3.61