Plot Model Predictions with Error Bars for Univariable GLM

S3 method to enable ggplot() in package ggplot2 to plot "glm_plotdata" objects ouptut by glm_plotdata().

Usage

# S3 method for class 'glm_plotdata'
ggplot(
  data = NULL,
  mapping = aes(),
  as_percent = FALSE,
  rev_y = FALSE,
  ...,
  environment = parent.frame()
)

Arguments

data

a data frame, or a data frame extension (e.g. a tibble).

mapping

Default list of aesthetic mappings to use for plot. If not specified, must be supplied in each layer added to the plot.

as_percent

logical. If TRUE, the y-axis uses a percentage scale; default FALSE.

rev_y

logical. If TRUE, the direction of the y-axis is reversed, which may be useful when plotting linear predictors; default FALSE.

...

further arguments passed to or from other methods. Not currently used.

environment

Used prior to tidy evaluation.

Value

A ggplot object.

Details

This S3 method plots model predictions and error bars representing confidence intervals or standard errors for a univariable glm with a categorical independent variable, optionally allowing representation of groupings of levels of the independent variable and faceting of a number of such plots.

ggplot.glm_plotdata() recognises a factor or character column in data named grouped for plotting grouped levels of an independent variable that are grouped within the underlying model. If levels are indeed grouped in the model, the data bars will be plotted with colour-coded borders representing the groups, and the ungrouped observed values contained in the data column level are plotted as symbols. If ungrouped levels are to be plotted, the grouped column should only contain NA values

A character column in data containing names of independent variables to be used for faceting may be identified by setting an attribute "facet_by" in data. Names of variables to be used for faceting may be converted to more informative facet labels by using a user-defined, vectorised labeller() function which should be named var_labs(), see labeller under facet_wrap().

If an individual plot, rather than a faceted plot, is printed, the name of the independent variable, converted by var_labs() (if provided), will be used for the plot title. The plot title, subtitle and axis labels may be overridden using the usual ggplot() syntax, see examples.

See also

facet_wrap(), ggplot(), labeller().

Other plot_model: glm_plotdata(), glm_plotlist(), var_labs()

Examples

## Example uses randomly generated data; re-running may be worthwhile.  

oldtheme <- theme_get()   ## Save ggplot defaults for later restoration

## Set ggplot defaults for pretty printing
theme_update(
  plot.title = element_text(color = "black", size = 20, hjust = 0.5),
  plot.subtitle = element_text(color = "black", size = 18, hjust = 0.5),
  axis.text.x = element_text(color = "black", size = 15),
  axis.text.y = element_text(color = "black", size = 15),
  axis.title.x = element_text(color = "black", size = 15),
  axis.title.y = element_text(color = "black", size = 15),
  strip.text.x = element_text(color = "black", size = 15),
  legend.position = "none"
)

## "labeller()" function to provide plot titles - see var_labs()
var_labs <- as_labeller(
    c(iv = "Risk Factor (Ungrouped Levels)",
      iv2 = "Risk Factor (Grouped Levels)")
)

## Create binomial data with groupings
(d <- list(iv2 = list(ab = c("a", "b"), cd = c("c", "d"))) |>
    add_grps(binom_data(), iv, .key = _))
#> __________________________
#> Simulated Binomial Data: -
#> 
#> # A tibble: 5 × 4
#>   iv    iv2      pn    qn
#>   <fct> <fct> <int> <int>
#> 1 a     ab       30    36
#> 2 b     ab       29    37
#> 3 c     cd       19    47
#> 4 d     cd       12    54
#> 5 e     e         4    62

## Ungrouped GLM plot data on linear predictor scale
(dp <- glm_plotdata(d, .dep_var = cbind(pn, qn), .ind_var = iv))
#> ________________
#> GLM Plot Data: -
#> 
#> # A tibble: 5 × 7
#>   level grouped     n    obs   pred  lower  upper
#> * <fct> <fct>   <int>  <dbl>  <dbl>  <dbl>  <dbl>
#> 1 a     NA         66 -0.182 -0.182 -0.669  0.304
#> 2 b     NA         66 -0.244 -0.244 -0.732  0.244
#> 3 c     NA         66 -0.906 -0.906 -1.44  -0.371
#> 4 d     NA         66 -1.50  -1.50  -2.13  -0.876
#> 5 e     NA         66 -2.74  -2.74  -3.76  -1.73 

## Plot model predictions and CI error bars
dp |> ggplot()


## Plot model predictions and CI error bars with reversed y-axis
dp |> ggplot(rev_y = TRUE)


## Grouped GLM plot data on linear predictor scale
(dp <- glm_plotdata(d, .dep_var = cbind(pn, qn), .ind_var = iv2, .ungroup = iv))
#> ________________
#> GLM Plot Data: -
#> 
#> # A tibble: 5 × 7
#>   level grouped     n    obs   pred  lower  upper
#> * <fct> <fct>   <int>  <dbl>  <dbl>  <dbl>  <dbl>
#> 1 a     ab         66 -0.182 -0.213 -0.557  0.131
#> 2 b     ab         66 -0.244 -0.213 -0.557  0.131
#> 3 c     cd         66 -0.906 -1.18  -1.59  -0.777
#> 4 d     cd         66 -1.50  -1.18  -1.59  -0.777
#> 5 e     e          66 -2.74  -2.74  -3.76  -1.73 

## Plot model predictions and CI error bars with reversed y-axis
dp |> ggplot(rev_y = TRUE)


## Ungrouped GLM plot data on reponse scale
(dp <- glm_plotdata(d, .dep_var = cbind(pn, qn), .ind_var = iv, type = "response"))
#> ________________
#> GLM Plot Data: -
#> 
#> # A tibble: 5 × 7
#>   level grouped     n    obs   pred  lower upper
#> * <fct> <fct>   <int>  <dbl>  <dbl>  <dbl> <dbl>
#> 1 a     NA         66 0.455  0.455  0.339  0.575
#> 2 b     NA         66 0.439  0.439  0.325  0.561
#> 3 c     NA         66 0.288  0.288  0.191  0.408
#> 4 d     NA         66 0.182  0.182  0.106  0.294
#> 5 e     NA         66 0.0606 0.0606 0.0229 0.151

## Plot model predictions and CI error bars
dp |> ggplot()


## Plot model predictions and CI error bars, with y-axis as percentage
dp |> ggplot(as_percent = TRUE)


## Grouped GLM plot data on reponse scale
(dp <- glm_plotdata(d, .dep_var = cbind(pn, qn), .ind_var = iv2, .ungroup = iv, type = "response"))
#> ________________
#> GLM Plot Data: -
#> 
#> # A tibble: 5 × 7
#>   level grouped     n    obs   pred  lower upper
#> * <fct> <fct>   <int>  <dbl>  <dbl>  <dbl> <dbl>
#> 1 a     ab         66 0.455  0.447  0.364  0.533
#> 2 b     ab         66 0.439  0.447  0.364  0.533
#> 3 c     cd         66 0.288  0.235  0.170  0.315
#> 4 d     cd         66 0.182  0.235  0.170  0.315
#> 5 e     e          66 0.0606 0.0606 0.0229 0.151

## Plot model predictions and CI error bars
dp |> ggplot(as_percent = TRUE)


## Grouped GLM plot data on reponse scale with standard errors 
(dp <- glm_plotdata(
                    d, .dep_var = cbind(pn, qn), .ind_var = iv2,
                    .ungroup = iv, conf_level = NA, type = "response"
                   ))
#> ________________
#> GLM Plot Data: -
#> 
#> # A tibble: 5 × 7
#>   level grouped     n    obs   pred  lower  upper
#> * <fct> <fct>   <int>  <dbl>  <dbl>  <dbl>  <dbl>
#> 1 a     ab         66 0.455  0.447  0.404  0.491 
#> 2 b     ab         66 0.439  0.447  0.404  0.491 
#> 3 c     cd         66 0.288  0.235  0.200  0.274 
#> 4 d     cd         66 0.182  0.235  0.200  0.274 
#> 5 e     e          66 0.0606 0.0606 0.0371 0.0975

## Plot model predictions and standard error bars
dp |> ggplot(as_percent = TRUE)


## Add x-axis label and bespoke titles
dp |> ggplot(as_percent = TRUE) +
labs(
    x = "Level",
    title = "Example for ggplot.glm_plotdata()",
    subtitle = "Fascinating Results"
)


theme_set(oldtheme)    ## Restore original ggplot defaults
rm(d, dp, oldtheme)