R中ggplot2水平堆叠条形图的变量与标签排序问题求助

核心需求回顾 #

需要绘制水平堆叠条形图，实现：

• 年龄组从最小(0: <18)到最大(5: 35+)排序，最小年龄组显示在条形最左侧
• 图例顺序正确，颜色与对应年龄组匹配
• 数值标签与年龄组对应准确

修正后的完整代码 #

# 1. 正确设置因子水平：按年龄从小到大排序，同时配置显示标签
bospop_by_agegroup_all_years$agegroup_redef <- factor(
  bospop_by_agegroup_all_years$agegroup_redef,
  levels = c("<18", "18-19", "20 to 24", "25-29", "30-34", "35+"),
  labels = c("0: <18", "1: 18-19", "2: 20 to 24", "3: 25-29", "4: 30-34", "5: 35+")
)

# 若使用sjlabelled管理标签，替换上述代码为：
# bospop_by_agegroup_all_years$agegroup_redef <- factor(
#   bospop_by_agegroup_all_years$agegroup_redef,
#   levels = c("<18", "18-19", "20 to 24", "25-29", "30-34", "35+")
# )
# bospop_by_agegroup_all_years$agegroup_redef <- set_labels(
#   bospop_by_agegroup_all_years$agegroup_redef,
#   labels = c("<18" = "0: <18", "18-19" = "1: 18-19", "20 to 24" = "2: 20 to 24", 
#              "25-29" = "3: 25-29", "30-34" = "4: 30-34", "35+" = "5: 35+")
# )

# 2. 绘制图形
ggplot(bospop_by_agegroup_all_years, aes(fill = agegroup_redef, y = N, x = year, label = label_numberformat(N))) +
  geom_bar(position = position_stack(reverse = FALSE), stat = "identity") +
  scale_y_continuous(n.breaks = 7, minor_breaks = seq(0, 700000, 50000), labels = label_numberformat) +
  scale_x_discrete(labels = get_labels(bospop_by_agegroup_all_years$year)) +
  # 若原调色板palette_households是按年龄从大到小配置的，需改为rev(palette_households)
  scale_fill_manual(values = palette_households, labels = get_labels(bospop_by_agegroup_all_years$agegroup_redef)) +
  labs(x = "", y = "", fill = "Age Group", title = "Boston's Population by Age Group, 1980-2023") +
  theme(
    legend.position = "bottom", 
    panel.background = element_blank(),
    panel.grid.major.x = element_line(color = "gray72", linewidth = 0.5, linetype = 1),
    panel.grid.minor.x = element_line(color = "gray72", linewidth = 0.5, linetype = 2),
    text = element_text(family = "Open Sans"),
    title = element_text(family = "Open Sans Semibold", size = 20),
    legend.title = element_text(family = "Open Sans Semibold", size = 14),
    legend.text = element_text(size = 12),
    legend.box.margin = margin(0, 0, 0, 0),
    axis.ticks.length = unit(0,  "lines"),
    axis.text.y = element_text(size = 12, margin = margin(0, -40, 0, 0)),
    axis.text.x = element_text(size = 12, margin = margin(0, 0, 0,0), family = "Open Sans Semibold", color = "black")
  ) +
  geom_text(position = position_stack(reverse = FALSE, vjust = 0.5),
            color = "black",
            size = 3.85,
            family = "Open Sans Semibold"
  ) + 
  coord_flip()

之前方案的问题分析 #

1. 选项1：因子水平按年龄从大到小定义，导致堆叠顺序反向（最左侧为35+），虽然图例标签正确，但堆叠逻辑不符合需求。
2. 选项2：因子水平顺序正确，但scale_fill_manual的调色板palette_households顺序与因子水平不匹配，造成图例颜色和条形颜色对应错误。
3. 选项3：用rev(agegroup_redef)反转因子水平，堆叠顺序变回从大到小，违背需求。

关键注意事项 #

1. 因子水平是核心逻辑：ggplot的堆叠顺序、图例顺序完全由因子levels的顺序决定，必须先按年龄从小到大定义levels。
2. coord_flip的视觉逻辑：水平条形的左侧对应原y轴的底部，因此因子levels的第一个元素（<18）会自动显示在最左侧。
3. 颜色与标签的匹配：若原调色板是按年龄从大到小配置的，需用rev(palette_households)反转颜色顺序，确保颜色与图例标签一一对应。

内容的提问来源于stack exchange，提问作者user22460492