ggplot柱状图误差线对齐、图例优化及堆叠图实现问题

1. 误差线与柱状图对齐修复 #

误差线错位的核心原因是：geom_errorbar没有继承pattern=treated的分组逻辑，且stat_summary的使用冗余（你的数据已是均值，无需再计算均值）。修正步骤：

• 将pattern=treated放入全局aes，让柱状图和误差线共用同一分组规则
• 统一position_dodge的宽度参数（保持0.9），确保两者对齐
• 替换stat_summary为geom_bar_pattern(stat="identity")，直接使用现有数据

2. 图例优化 #

当前图例无明确说明，需给scale_pattern_manual添加名称和对应标签，同时隐藏多余的填充图例：

• 设置图例名称为「处理状态」，明确对应「未处理/处理」的图案样式
• 用scale_fill_identity直接使用数据中的颜色值，避免生成无效的填充图例

3. 分面内堆叠柱状图实现 #

堆叠需要将position改为position_stack()，同时调整x轴分组为models（同一x类别下堆叠不同处理组），误差线需用position_stack(vjust=0.5)居中显示在堆叠段上方。

修正后的分组柱状图代码（对齐+图例优化） #

library(ggplot2)
library(tidyverse)
library(ggpattern)

models = c("a", "b")
task = c("1","2")
ratios = c(0.3, 0.4)
standard_errors = c(0.02, 0.02)
colors = c("#F39B7FFF", "#8491B4FF")

# 整理数据（简化流程）
df <- data.frame(task = task, ratios = ratios) %>%
  mutate(filler = 1 - ratios) %>%
  gather(key = "obs", value = "ratios", -1) %>%
  mutate(upper = ratios + rep(standard_errors, 2),
         lower = ratios - rep(standard_errors, 2),
         models = rep(models, 2),
         col = rep(colors, 2),
         # 明确处理组标签
         treated = ifelse(ratios < 0.5, "未处理（not treated）", "处理（treated）"))

# 对齐后的分组柱状图
p_aligned <- ggplot(df, aes(x = models, y = ratios, fill = col, 
                            pattern = treated, ymin = lower, ymax = upper)) +
  geom_bar_pattern(stat = "identity", position = position_dodge(width = 0.9),
                   pattern_fill = "black", colour = "black") +
  geom_errorbar(position = position_dodge(width = 0.9), width = 0.2) +
  scale_pattern_manual(name = "处理状态", 
                       values = c("未处理（not treated）" = "none", "处理（treated）" = "stripe")) +
  scale_fill_identity() + # 直接使用数据中的颜色
  facet_grid(.~task, scales = "free_x", space = "free_x", switch = "x") +
  labs(x = "模型", y = "比例")
p_aligned

分面内堆叠柱状图代码 #

# 堆叠版本
p_stacked <- ggplot(df, aes(x = models, y = ratios, fill = treated, pattern = treated)) +
  geom_bar_pattern(stat = "identity", position = position_stack(),
                   pattern_fill = "black", colour = "black") +
  # 误差线居中显示在堆叠段上方
  geom_errorbar(aes(ymin = lower, ymax = upper), 
                position = position_stack(vjust = 0.5), width = 0.2) +
  scale_pattern_manual(name = "处理状态", 
                       values = c("未处理（not treated）" = "none", "处理（treated）" = "stripe")) +
  scale_fill_manual(name = "处理状态",
                    values = c("未处理（not treated）" = "#F39B7FFF", "处理（treated）" = "#8491B4FF")) +
  facet_grid(.~task, scales = "free_x", space = "free_x", switch = "x") +
  labs(x = "模型", y = "比例")
p_stacked

内容的提问来源于stack exchange，提问作者Tamay