具有多个几何图形的ggplot图例顺序答案

【问题标题】：ggplot legend order with multiple geoms具有多个几何图形的ggplot图例顺序
【发布时间】：2020-06-24 15:35:53
【问题描述】：

library(tidyverse)
df <- tibble(col1 = as.Date(c("2020-01-01", "2020-01-01", "2020-01-01", 
                              "2020-02-01", "2020-02-01", "2020-02-01")),
             col2 = factor(rep(c("A", "B", "C"), 2), levels = c("C", "A", "B")),
             col3 = c(8, 3, 2, 9, 1, 5))

df
#> # A tibble: 6 x 3
#>   col1       col2   col3
#>   <date>     <fct> <dbl>
#> 1 2020-01-01 A         8
#> 2 2020-01-01 B         3
#> 3 2020-01-01 C         2
#> 4 2020-02-01 A         9
#> 5 2020-02-01 B         1
#> 6 2020-02-01 C         5

levels(df$col2)
#> [1] "C" "A" "B"

我已经创建了数据框df 和col2，如上所示。我认为这种因素的顺序决定了 ggplot 图例条目的默认顺序，但事实并非如此，如下所示。

ggplot(df, aes(col1, col3, fill = col2)) + 
  geom_col(data = df %>% filter(col2 != "C"), 
           position = "dodge", 
           color = "black") + 
  geom_col(data = df %>% filter(col2 == "C"), 
           width = 15,
           color = "black") +
  guides(fill = guide_legend(title = "Legend"))

我将推测 col2 级别排序 (#> [1] "C" "A" "B") 在 ggplot 的图例排序中不受尊重，因为有多个我的代码中的几何图形。它是否正确？如何让图例条目按以下顺序出现？

C
一个
B

【问题讨论】：

标签： r ggplot2

【解决方案1】：

您可以在色标上设置限制：

library(tidyverse)
df <- tibble(col1 = as.Date(c("2020-01-01", "2020-01-01", "2020-01-01", 
                              "2020-02-01", "2020-02-01", "2020-02-01")),
             col2 = factor(rep(c("A", "B", "C"), 2), levels = c("C", "A", "B")),
             col3 = c(8, 3, 2, 9, 1, 5))

ggplot(df, aes(col1, col3, fill = col2)) + 
  geom_col(data = df %>% filter(col2 != "C"), 
           position = "dodge", 
           color = "black") + 
  geom_col(data = df %>% filter(col2 == "C"), 
           width = 15,
           color = "black") +
  scale_fill_discrete(name = "Legend",
                      limits = c("C", "A", "B"))

或者，您可以在比例中设置drop = FALSE，这样它就不会在层之间丢弃未使用的级别：

ggplot(df, aes(col1, col3, fill = col2)) + 
  geom_col(data = df %>% filter(col2 != "C"), 
           position = "dodge", 
           color = "black") + 
  geom_col(data = df %>% filter(col2 == "C"), 
           width = 15,
           color = "black") +
  scale_fill_discrete(name = "Legend", drop = FALSE)

【讨论】：

【解决方案2】：

只需设置 scale_fill_discrete 中断参数...

library(tidyverse)

df <- tibble(col1 = as.Date(c("2020-01-01", "2020-01-01", "2020-01-01", 
                              "2020-02-01", "2020-02-01", "2020-02-01")),
             col2 = factor(rep(c("A", "B", "C"), 2), levels = c("C", "A", "B")),
             col3 = c(8, 3, 2, 9, 1, 5))


ggplot(df, aes(col1, col3, fill = col2)) + 
  geom_col(data = df %>% filter(col2 != "C"), 
           position = "dodge", 
           color = "black") + 
  geom_col(data = df %>% filter(col2 == "C"), 
           width = 15,
           color = "black") +
  scale_fill_discrete(breaks = c("C", "A", "B"))+
  guides(fill = guide_legend(title = "Legend"))

^{由reprex package (v0.3.0) 于 2020 年 6 月 24 日创建}

【讨论】：