Skip to contents

Graphical options

Source: vignettes/articles/options.qmd

visreg tries to set up pleasant-looking default options, but everything can be tailored to user specifications. For the plots below, we work from this general model:

airquality$Heat <- cut(airquality$Temp, 3, labels = c("Cool", "Mild", "Hot"))
fit <- lm(Ozone ~ Solar.R + Wind + Heat, data = airquality)

Turning on/off plot components

By default, visreg includes the fitted line, confidence bands, and partial residuals, but the residuals and the bands can be turned off:

visreg(fit, "Wind", band = FALSE)

visreg(fit, "Wind", partial = FALSE)

visreg(fit, "Wind", partial = FALSE, rug = TRUE)

Jittering

If there are many ties in a numeric variable x, jittering can be helpful way to avoid overplotting. Compare the following two plots:

month_fit <- lm(Ozone ~ Solar.R + Wind + poly(Month, 2), data = airquality)
visreg(month_fit, "Month")
visreg(month_fit, "Month", jitter = TRUE)

Appearance of points, lines, and bands

Specifying col='red' won’t work, because visreg can’t know whether you’re trying to change the color of the line, the band, or the points. These options must be specified through separate parameter lists:

  • line: Controls the appearance of the fitted line
  • fill: Controls the appearance of the confidence band
  • points: Controls the appearance of the partial residuals

Each of these is passed along to the underlying ggplot2 geom (geom_line/geom_ribbon/geom_point for a continuous x, or geom_errorbar/geom_tile/geom_jitter for a factor x), so any parameter that geom accepts can be specified:

visreg(
  fit,
  "Wind",
  line = list(color = "red"),
  fill = list(fill = "green"),
  points = list(size = 2, shape = 1)
)

Generic plot options

visreg returns a gg object, so its appearance can be customized just like any other ggplot2 plot, by adding further components. Here’s an example that adds a title and re-labels the (log-transformed) y-axis:

fit <- lm(log(Ozone) ~ Solar.R + Wind + Temp, data = airquality)
at <- seq(1.5, 5, 0.5)
visreg(fit, "Wind") +
  labs(title = "Ozone is bad for you", y = "Ozone") +
  scale_y_continuous(breaks = at, labels = round(exp(at), 1))

Band width for factors

When x is a factor, the width of the confidence band (and its accompanying line) is controlled through the fill/line arguments’ width parameter:

fit <- lm(Ozone ~ Solar.R + Wind + Heat, data = airquality)
visreg(fit, "Heat", line = list(width = 0.98), fill = list(width = 0.98))
visreg(fit, "Heat", line = list(width = 0.1), fill = list(width = 0.1))

Subsetting the plot

Occasionally, you might want to plot only a subset of the levels or observations; you can use subset to accomplish this:

v <- visreg(fit, "Wind", by = "Heat", plot = FALSE)
v1 <- subset(v, Heat %in% c("Cool", "Hot"))
plot(v1)

v2 <- subset(v, Wind < 15)
plot(v2)