ggpairs errors out if there is missing data in an ordered factor variable

[janlimbeck]

ggpairs seems to error out when applied to a data set that has missing data in an ordered factor variable. ggpairs works fine if there is missing data in an unordered factor variable. Tests were carried out with the latest package versions on cran as of Febuary 1st 2023.

library(GGally)

# example using the iris data set
df <- iris

# ggairs works as expected
ggpairs(df)

# turn species column into an ordered factor
df$Species <- factor(df$Species, levels = levels(df$Species), ordered = T)

# ggairs works still as expected
ggpairs(df)

# simulate missing data
df[1, 'Species'] <- NA

# ggpairs errors out when there is missing data in an ordered factor variable
ggpairs(df)

df <- iris

# simulate missing data
df[1, 'Species'] <- NA

# ggpairs works correctly when there is missing data in an un-ordered factor variable
ggpairs(df)

The error message I am getting is below

Error:
! Can't combine `..1$Species` <factor<fb977>> and `..2$Species` <ordered<fb977>>.
---
Backtrace:
  1. base (local) `<fn>`(x)
 31. ggplot2 (local) `<fn>`(`<vctrs__2>`)
 32. vctrs::vec_ptype2(x, y, x_arg = x_arg, y_arg = y_arg, call = call)
 33. vctrs (local) `<fn>`()
 34. vctrs:::vec_ptype2.factor.ordered(...)
 35. vctrs::vec_default_ptype2(x, y, ...)
 39. vctrs::stop_incompatible_type(...)
 40. vctrs:::stop_incompatible(...)
 41. vctrs:::stop_vctrs(...)