如何在ggplot中按分组（而非条件）为分组条形图着色？

你需要创建分组条形图，x轴为decade，y轴为rate，按condition分隔条形，且不同decade使用不同颜色。之前的代码因映射逻辑问题未达到预期效果，以下是几种可行的解决方案：

方法1：用填充颜色区分年代，线条样式区分条件 #

这种方法让同一年代的两个条形共享主色调，通过线条样式区分Yes/No条件，视觉上清晰区分年代和条件：

library(ggplot2)

df <- data.frame (decade  = c("1970", "1970","1980", "1980","1990","1990","2000","2000"),
                  condition = c("Yes", "No", "Yes","No","Yes","No","Yes","No"),
                  rate = c(60,34,50,25,24,12,9,5))

ggplot(df, aes(x = decade, y = rate, fill = decade, linetype = condition)) +
  geom_bar(position = "dodge", stat = "identity", color = "black") +
  scale_fill_brewer(palette = "Set1") + # 可选：使用预定义配色方案
  labs(x = "年代", y = "比率", fill = "年代", linetype = "条件") +
  theme_minimal()

逻辑说明： #

• 将fill映射为decade，确保同一年代的条形颜色一致
• 用linetype映射condition，通过实线/虚线区分Yes/No
• 添加color="black"是为了让线条样式更明显

方法2：自定义同色系深浅区分条件，年代用不同主色 #

如果希望同一年代的两个条件有细微颜色差异，可通过交互变量自定义配色：

library(ggplot2)

# 自定义配色：每个年代对应深浅两种颜色
decade_colors <- c(
  "1970_Yes" = "#1f77b4", "1970_No" = "#aec7e8",
  "1980_Yes" = "#ff7f0e", "1980_No" = "#ffbb78",
  "1990_Yes" = "#2ca02c", "1990_No" = "#98df8a",
  "2000_Yes" = "#d62728", "2000_No" = "#ff9896"
)

ggplot(df, aes(x = decade, y = rate, fill = interaction(decade, condition))) +
  geom_bar(position = "dodge", stat = "identity") +
  scale_fill_manual(
    values = decade_colors,
    labels = c("1970-Yes", "1970-No", "1980-Yes", "1980-No", 
               "1990-Yes", "1990-No", "2000-Yes", "2000-No"),
    name = "分组"
  ) +
  labs(x = "年代", y = "比率") +
  theme_minimal()

逻辑说明： #

• 使用interaction(decade, condition)创建组合分组变量
• 手动为每个组合分配同色系的深浅颜色，既区分年代，又区分条件

方法3：用图案区分条件（需ggpattern包） #

如果希望保持年代颜色完全一致，用图案区分条件，可借助ggpattern包：

# 先安装包：install.packages("ggpattern")
library(ggplot2)
library(ggpattern)

ggplot(df, aes(x = decade, y = rate, fill = decade, pattern = condition)) +
  geom_bar_pattern(
    position = "dodge", stat = "identity",
    pattern_fill = "black", pattern_density = 0.1
  ) +
  scale_fill_brewer(palette = "Set1") +
  scale_pattern_manual(values = c(Yes = "none", No = "stripe")) +
  labs(x = "年代", y = "比率", fill = "年代", pattern = "条件") +
  theme_minimal()

逻辑说明： #

• fill映射decade保证同年代颜色一致
• pattern映射condition，用无图案/条纹区分Yes/No

原代码问题分析 #

你之前的代码中fill=condition将填充颜色绑定到了条件，color=decade仅设置条形边框颜色，因此无法看到年代对应的填充效果。调整美学映射的绑定对象，就能实现需求。

内容的提问来源于stack exchange，提问作者Hong