Diverging color scale for Pearson residuals

Provides a red-white-blue color scale centered at 0 for visualizing Pearson residuals from loglinear models. Designed for use with geom_mosaic() when a model is supplied through the layer's expected argument or through mosaic_settings.

Usage

scale_fill_residual(
  ...,
  low = "darkred",
  mid = "white",
  high = "darkblue",
  midpoint = 0,
  limits = NULL,
  name = "Pearson\nResidual"
)

scale_fill_residuals(
  ...,
  low = "darkred",
  mid = "white",
  high = "darkblue",
  midpoint = 0,
  limits = NULL,
  name = "Pearson\nResidual"
)

Arguments

...

Arguments passed to scale_fill_gradient2

low

Color for negative residuals (default: "darkred")

mid

Color for zero residuals (default: "white")

high

Color for positive residuals (default: "darkblue")

midpoint

Center point for color scale (default: 0)

limits

Range used for the color gradient. Values beyond supplied limits receive the corresponding endpoint color.

name

Legend title

Value

A ggplot2 fill scale that can be added to a plot.

Details

The default legend always labels -4, 0, and 4. It also labels supplied limits and the observed minimum and maximum when those differ from the limits. The legend extends to every labeled value, with solid endpoint color beyond supplied limits. When the contributing mosaic cells have outlines, positive residuals have a solid dark blue outline, negative residuals have a dashed dark red outline, and an unoutlined midpoint band (white by default) separates them at zero. Setting colour = NA on every contributing mosaic layer removes these outlines from both the cells and the legend. Black ticks are drawn outside the color bar, which stretches with the mosaic panel. Nearby vertical labels are separated, and a thin elbow connects each displaced label to its exact tick. The neighboring label uses a longer straight tick so nearby text shares a common alignment. Automatically generated numeric labels are rounded to one decimal place. The legend can be hidden normally with theme(legend.position = "none").

Author

Gavin Klorfine

Examples

data(titanic)

# Independence model with residual shading
ggplot(data = titanic, aes(x = product(Class, Sex))) +
  geom_mosaic(expected = "independence") +
  scale_fill_residual()


# Custom colors
ggplot(data = titanic, aes(x = product(Class, Sex))) +
  geom_mosaic(expected = "independence") +
  scale_fill_residual(low = "red", high = "blue")


# Custom limits to highlight strong deviations
ggplot(data = titanic, aes(x = product(Class, Sex, Survived))) +
  geom_mosaic(expected = ~ Class + Sex) +
  scale_fill_residual(limits = c(-4, 4))