Plot a timeseries of relative scores versus time

Usage

plot_rel_score_t(
  scores,
  target_models,
  baseline_model,
  metric_to_compare,
  x = "forecast_date",
  by = NULL,
  retain_baseline = FALSE,
  model_z_order = NULL
)

Arguments

scores: Output of scoringutils::score(), not yet summarized.
target_models: Model for which to plot relative scores
baseline_model: Baseline model for the relative scores computation.
metric_to_compare: Metric for which to plot relative scores. One of "wis" or "crps". Not case-sensitive.
x: Name of the column defining the x axis (typically time) Default "forecast_date".
by: additional variables defining a single score to plot. Passed along with the variable defined by x as the by argument to forecasttools::summarise_scores_with_baseline(). Default NULL (do not summarize over additional variables).
retain_baseline: Retain the baseline model (with its relative scores of 1) alongside the target models(s)? Default FALSE.
model_z_order: z-order in which to overplot the individual models, ascending (so the last named model is plotted on top). If NULL (default), plot the the models in order of overall score, so that the lowest (best) scoring models are on top.

Value

a ggplot object with the distributions plotted by forecast date.