plot_pairwise_comparison: Plot Heatmap of Pairwise Comparisons

In scoringutils: Utilities for Scoring and Assessing Predictions

Plot Heatmap of Pairwise Comparisons

Description

Creates a heatmap of the ratios or pvalues from a pairwise comparison between models

Usage

plot_pairwise_comparison(
  comparison_result,
  type = c("mean_scores_ratio", "pval", "together"),
  smaller_is_good = TRUE
)

Arguments

comparison_result	A data.frame as produced by pairwise_comparison()
type	character vector of length one that is either "mean_scores_ratio", "pval", or "together". This denotes whether to visualise the ratio or the p-value of the pairwise comparison or both. Default is "mean_scores_ratio".
smaller_is_good	logical (default is TRUE) that indicates whether smaller or larger values are to be interpreted as 'good' (as you could just invert the mean scores ratio). This option is not supported when type = "pval"

Examples

library(ggplot2)
df <- data.frame(
  model = rep(c("model1", "model2", "model3"), each = 10),
  id = rep(1:10),
  interval_score = abs(rnorm(30, mean = rep(c(1, 1.3, 2), each = 10))),
  ae_median = (abs(rnorm(30)))
)

scores <- score(example_quantile)
pairwise <- pairwise_comparison(scores, by = "target_type")
plot_pairwise_comparison(pairwise) +
  facet_wrap(~target_type)

scoringutils documentation built on May 14, 2022, 1:06 a.m.