在ggplot2中使用表达式作图例标签时如何合并线型与颜色图例？

当使用表达式（如希腊符号）作为图例标签时，ggplot2会因标签类型不匹配拆分原本可合并的线型与颜色图例，以下是具体解决方法：

问题复现 #

先看导致图例拆分的典型代码：

library(ggplot2)

set.seed(123) # 固定随机种子保证结果可复现
data <- data.frame(Train = rep(c(20,30,40,50),2), xi = runif(8,0,1), category = c(rep("0.3_val",4),rep("0.3_test",4)))
data <- rbind(data, data.frame(Train = rep(c(20,30,40,50),2), xi = runif(8,0,1), category = c(rep("0.6_val",4),rep("0.6_test",4))))

# 此代码会拆分图例：color用表达式标签，linetype用普通字符标签
ggplot(data, aes(x = Train, y = xi, color = category, linetype = category)) +
  geom_line() +
  scale_color_manual(
    values = c("0.3_val" = "red", "0.3_test" = "blue", "0.6_val" = "darkred", "0.6_test" = "darkblue"),
    labels = c(expression(paste(xi, "=0.3 (训练)")), expression(paste(xi, "=0.3 (测试)")),
               expression(paste(xi, "=0.6 (训练)")), expression(paste(xi, "=0.6 (测试)")))
  ) +
  scale_linetype_manual(
    values = c("0.3_val" = 1, "0.3_test" = 2, "0.6_val" = 1, "0.6_test" = 2),
    labels = c("xi=0.3 (训练)", "xi=0.3 (测试)", "xi=0.6 (训练)", "xi=0.6 (测试)")
  )

解决方法 #

核心是让scale_color_manual和scale_linetype_manual的图例标题、标签内容与类型完全一致，强制ggplot2合并图例：

修正后的代码 #

library(ggplot2)

set.seed(123)
data <- data.frame(Train = rep(c(20,30,40,50),2), xi = runif(8,0,1), category = c(rep("0.3_val",4),rep("0.3_test",4)))
data <- rbind(data, data.frame(Train = rep(c(20,30,40,50),2), xi = runif(8,0,1), category = c(rep("0.6_val",4),rep("0.6_test",4))))

# 定义统一的表达式标签
legend_labels <- c(
  expression(paste(xi, "=0.3 (训练)")),
  expression(paste(xi, "=0.3 (测试)")),
  expression(paste(xi, "=0.6 (训练)")),
  expression(paste(xi, "=0.6 (测试)"))
)

ggplot(data, aes(x = Train, y = xi, color = category, linetype = category)) +
  geom_line(linewidth = 1) +
  scale_color_manual(
    name = "参数与数据集", # 统一图例标题
    values = c("0.3_val" = "red", "0.3_test" = "blue", "0.6_val" = "darkred", "0.6_test" = "darkblue"),
    labels = legend_labels,
    guide = guide_legend(override.aes = list(linewidth = 1.2)) # 可选：调整图例线条粗细
  ) +
  scale_linetype_manual(
    name = "参数与数据集", # 与color scale的标题完全一致
    values = c("0.3_val" = 1, "0.3_test" = 2, "0.6_val" = 1, "0.6_test" = 2),
    labels = legend_labels # 与color scale的标签完全一致
  ) +
  theme_minimal()

原理说明 #

ggplot2判断是否合并图例的依据是：两个美学映射（color/linetype）对应的scale是否具有相同的标题、标签内容及类型。只要保证这两个scale的name和labels完全匹配（包括都使用表达式类型），系统就会自动将它们合并为一个图例。

内容的提问来源于stack exchange，提问作者unknown_user