create.boxplot: How to keep all factor levels in plot, even when n=0 ?

[jarbet]

When making boxplots for a grouping variable, is there some way to keep all group levels in the plot, even if some of the levels have 0 patients?

The reason I want to do this is because I am using a for loop to make many plots, and sometimes n=0 for some of the group levels. But I dont want to drop the groups, since I want all of the plots to align the same way in a final multipanelplot.

Example

Notice the group level other is dropped in the final plot because n=0, but is there same way to keep a place for that group in the plot?

suppressPackageStartupMessages(library(BoutrosLab.plotting.general));

data(mtcars);
mtcars$am <- factor(
    x = mtcars$am,
    levels = 0:2,
    labels = c('automatic', 'manual', 'other')
    );

create.boxplot(
    formula = mpg ~ am,
    data = mtcars
    );

^{Created on 2024-05-28 with reprex v2.0.2}

[jarbet]

Ahh nvm, I think I found a solution:

You need to make a dummy sample with the missing factor level (and set all other variables to NA), then that group will appear in the plot:

suppressPackageStartupMessages(library(BoutrosLab.plotting.general));

data(mtcars);
mtcars$am <- factor(
    x = mtcars$am,
    levels = 0:2,
    labels = c('automatic', 'manual', 'other')
    );
dummy.sample <- data.frame(matrix(NA, nrow = 1, ncol = ncol(mtcars)));
colnames(dummy.sample) <- colnames(mtcars);
dummy.sample$am <- 'other';
mtcars <- rbind(mtcars, dummy.sample);

create.boxplot(
    formula = mpg ~ am,
    data = mtcars
    )

^{Created on 2024-05-28 with reprex v2.0.2}

Related to: #111