geom_point和coord_polar-未知美学警告

Question

I know geom_point doesn't have mapping for ymin and ymax, but when I don't use them it doesn't give the figure I expect. Nevertheless when using ymin and ymax I get the figure I wish, but I get warning of "Ignoring unknown aesthetics: ymin, ymax". I can't figure out what should be done to avoid warning but getting the plot I wish.

Here is example dataset:

ab <- data.frame(a = seq(50,100,by=10), b=LETTERS[1:6])

Without ymin and ymax but not the figure I expect:

library("ggplot2")

ggplot(ab) +
  scale_x_discrete() +
    scale_y_reverse() +
  geom_rect(xmin=Inf, xmax = -Inf, ymin = 0 - 50, ymax = 0, fill="lightblue") +
  geom_rect(xmin=Inf, xmax = -Inf, ymin = 0 - 82, ymax = 0 - 50, fill="lightgreen") +
  geom_rect(xmin=Inf, xmax = -Inf, ymin = -100, ymax = 0 - 80, fill="darkgreen") +
  geom_vline(xintercept=1:6, size=1.5, color="white") +
  geom_hline(yintercept=c(50, 80, 100), size=0.1, color="white") +
  geom_point(aes(x=b, y=a),
             shape=21, fill="#FF9933", size=10, position="identity") +
  ggtitle("Test") +
  coord_polar() +
  theme(
    plot.title=element_text(face = "bold", color = "black", size = 17, ),
    panel.background=element_rect(fill = c("white")),
    panel.grid=element_blank(),
    panel.grid.major=element_line(size=2),
    panel.grid.minor.y=element_blank(),
    axis.text.x=element_text(vjust=5),
    axis.text=element_text(size = 12, color = "black", face = "bold"),
    axis.text.y=element_blank(),
    axis.title.x=element_blank(),
    axis.title.y=element_blank(),
    axis.ticks=element_blank())

with ymin and ymax in geom_point (the expected output with warning):

ggplot(ab) +
  scale_x_discrete() +
    scale_y_reverse() +
  geom_rect(xmin=Inf, xmax = -Inf, ymin = 0 - 50, ymax = 0, fill="lightblue") +
  geom_rect(xmin=Inf, xmax = -Inf, ymin = 0 - 82, ymax = 0 - 50, fill="lightgreen") +
  geom_rect(xmin=Inf, xmax = -Inf, ymin = -100, ymax = 0 - 80, fill="darkgreen") +
  geom_vline(xintercept=1:6, size=1.5, color="white") +
  geom_hline(yintercept=c(50, 80, 100), size=0.1, color="white") +
  geom_point(aes(x=b, y=a, ymin = 0, ymax = 100),
             shape=21, fill="#FF9933", size=10, position="identity") +
  ggtitle("Test") +
  coord_polar() +
  theme(
    plot.title=element_text(face = "bold", color = "black", size = 17, ),
    panel.background=element_rect(fill = c("white")),
    panel.grid=element_blank(),
    panel.grid.major=element_line(size=2),
    panel.grid.minor.y=element_blank(),
    axis.text.x=element_text(vjust=5),
    axis.text=element_text(size = 12, color = "black", face = "bold"),
    axis.text.y=element_blank(),
    axis.title.x=element_blank(),
    axis.title.y=element_blank(),
    axis.ticks=element_blank())

Answer 1

In ggplot2, the data that you're plotting is what determines the plot area unless you explicitly change the plot limits. Your a variable (which you're plotting on the y-axis) only covers the range from 50 to 100. Thus, unless you specify a different y range to use in the plot, that's all you're seeing by default. In short, the proper way to fix this is to pass the limits argument to the scale_y_reverse call -- this tells the plot to plot a different range than the range the data covers. It will look like:

scale_y_reverse(limits=c(100, 0))

(The 100 comes first because you're reversing your y-axis.)

If you add that change to your plot without the ymin and ymax arguments, you'll get the graph you expect.

Now, if ymin and ymax aren't mapped, why are they changing your plot? I think the reason is that geometries (like geom_point ) don't remove aesthetic mappings that they don't understand. These mappings will get carried down to the layer, and if the layer understands the mappings, it will use them. That essentially "hacks" y-axis limits into the plot without specifying them the right way.