Skip to contents

Plotting

Source: vignettes/articles/plot.Rmd
plot.Rmd

erplots draws exposure-response plots from any model that implements the [er_model_interface]. This article uses logistic regression models fitted with erglm, but the plotting code itself has no knowledge of glm().

Fit the model first

Unlike a plotting function that fits a model behind the scenes, erplots expects you to fit the model yourself and pass it in explicitly:

mod <- erglm_model(ae1 ~ aucss, erglm_data, family = binomial())

Defining plots

Basic usage

erglm_data |> 
  er_plot(exposure = aucss, response = ae1) |> 
  er_plot_show_model(mod) |> 
  er_plot_show_quantiles() |> 
  plot()

Adding extra components

erglm_data |> 
  er_plot(exposure = aucss, response = ae1) |> 
  er_plot_show_model(mod) |> 
  er_plot_show_quantiles() |>
  er_plot_show_datastrip() |>
  er_plot_show_groups(group_by = aucss) |> 
  plot()

Stratification

Stratification adds colour across all components. This requires a model that includes the stratification variable as a term:

mod_strat <- erglm_model(ae1 ~ aucss + sex, erglm_data, family = binomial())

erglm_data |> 
  er_plot(
    exposure = aucss, 
    response = ae1, 
    stratify_by = sex
  ) |> 
  er_plot_show_model(mod_strat) |> 
  er_plot_show_quantiles() |> 
  er_plot_show_datastrip() |>
  plot()

You can suppress stratification for specific components

erglm_data |> 
  er_plot(
    exposure = aucss, 
    response = ae1, 
    stratify_by = sex
  ) |> 
  # keep_strata = FALSE needs a model that doesn't include the
  # stratification variable, so we pass the un-stratified `mod` here
  er_plot_show_model(mod, keep_strata = FALSE) |> 
  er_plot_show_quantiles() |> 
  er_plot_show_datastrip() |>
  plot()

Model component

Default is style = "ribbonline" but you can also draw spaghetti plots to represent parameter uncertainty. Spaghetti plots require the model to implement er_simulate() (erglm’s models do); models that only implement er_predict() fall back to "ribbonline" with a message.

erglm_data |> 
  er_plot(aucss, ae1) |> 
  er_plot_show_model(mod, style = "spaghetti") |> 
  er_plot_show_quantiles() |> 
  plot()
#> Using seed = 4371
#> Warning in ggplot2::geom_path(data = sim, mapping = ggplot2::aes(x =
#> .data[[exposure$name]], : Ignoring unknown parameters: `fill`

Quantile component

You can modify the number of bins:

erglm_data |> 
  er_plot(aucss, ae1) |> 
  er_plot_show_model(mod) |> 
  er_plot_show_quantiles(bins = 6) |> 
  plot()

You can also modify the confidence level for the Clopper-Pearson interval (this empirical-summary layer currently assumes a binary response):

erglm_data |> 
  er_plot(aucss, ae1) |> 
  er_plot_show_model(mod) |> 
  er_plot_show_quantiles(bins = 6, conf_level = .8) |> 
  plot()

Strip component

Group component

Multiple grouping variables are allowed:

erglm_data |> 
  er_plot(aucss, ae1) |> 
  er_plot_show_model(mod) |> 
  er_plot_show_quantiles() |>
  er_plot_show_groups(group_by = c(aucss, sex)) |> 
  plot()

Stratification propagates to the group component:

erglm_data |> 
  er_plot(aucss, ae1, stratify_by = sex) |> 
  er_plot_show_model(mod_strat) |> 
  er_plot_show_quantiles() |>
  er_plot_show_groups(group_by = aucss) |> 
  plot()

The default is style = "boxplot" but you can also use violin plots:

erglm_data |> 
  er_plot(aucss, ae1) |> 
  er_plot_show_model(mod) |> 
  er_plot_show_quantiles() |>
  er_plot_show_groups(group_by = sex, style = "violin") |> 
  plot()