在ggplot2中按组排序Odds Ratio图的技术求助

问题分析 #

你当前的代码中变量始终按字母排序的核心原因是：facet_grid(labels~.)是基于labels变量的顺序来排列分面的，仅将groups转为因子只会影响图例和着色，无法改变分面的排列顺序。

解决方案 #

要实现按组排序，需调整labels变量的因子水平顺序，具体操作如下：

• 指定你期望的组排列优先级（例如Household → Demographics → Behaviour）
• 将groups转换为带指定顺序的因子，确保分组排序时遵循该优先级
• 按groups分组排序数据，再把labels设置为因子，其水平顺序即为排序后的顺序

修改后的完整代码 #

library(ggplot2)

# 数据定义
pfhist_df <- data.frame(labels = c("Travel time to hospital > 10h",
                                   "Travel time to hospital > 18h",
                                   "Travel time to hospital > 28h",
                                   "Ethnicity: Dusun",
                                   "Ethnicity: Rungus",
                                   "Ethnicity: Sungoi",
                                   "Wealth: quartile 2", 
                                   "Wealth: quartile 3",
                                   "Wealth: quartile 4",
                                   "Occupation: student", 
                                   "Previously diagnosed with malaria", 
                                   "Elevation: 13 - 30m", 
                                   "Elevation > 213m", 
                                   "Travel time to clinic >7h", 
                                   "Travel time to clinic >10h", 
                                   "Head of household educated", 
                                   "Spend more than 3h outside daily", 
                                   "Travel to forest"), 
                        groups = c("Household", 
                                   "Household",
                                   "Household", 
                                   "Demographics",
                                   "Demographics",
                                   "Demographics",
                                   "Demographics",
                                   "Demographics",
                                   "Demographics",
                                   "Demographics",
                                   "Behaviour", 
                                   "Household",
                                   "Household",
                                   "Household",
                                   "Household",
                                   "Demographics",
                                   "Behaviour",
                                   "Behaviour"),
                        boxOdds = c(1.34,
                                    1.58,
                                    2.33,
                                    1.89,
                                    1.57,
                                    1.46,
                                    0.8, 
                                    0.69,
                                    0.48,
                                    0.43, 
                                    1.6, 
                                    1.26, 
                                    0.54, 
                                    1.22, 
                                    1.32,
                                    0.79, 
                                    1.57, 
                                    1.4),
                        boxCILow = c(1.09, 
                                     1.27, 
                                     1.83, 
                                     1.43, 
                                     1.21, 
                                     1.06, 
                                     0.67, 
                                     0.56, 0.38, 0.28, 1.36, 1.03, 0.4, 1.01, 1.08, 0.72, 1.11, 1.12),
                        boxCIHigh = c(1.64, 1.96, 2.95, 2.49, 2.05, 2.29, 0.99, 0.85, 0.6, 0.65, 
                                      1.88, 1.54, 0.72, 1.48, 1.63, 0.87, 2.23, 1.75))

# 关键：调整因子顺序实现按组排序
# 1. 指定组的期望排列顺序
group_order <- c("Household", "Demographics", "Behaviour")
pfhist_df$groups <- factor(pfhist_df$groups, levels = group_order)

# 2. 按groups分组，保持组内原顺序（若要按OR值排序，可替换为order(pfhist_df$groups, pfhist_df$boxOdds)）
pfhist_df <- pfhist_df[order(pfhist_df$groups), ]

# 3. 将labels设置为因子，水平顺序为排序后的顺序
pfhist_df$labels <- factor(pfhist_df$labels, levels = pfhist_df$labels)

# 绘图代码（与原代码一致，无需修改）
ggplot(pfhist_df, aes(x = boxOdds, y = groups)) + 
  geom_errorbarh(aes(xmin = boxCILow, xmax = boxCIHigh), size = .5, height = .2, color = "gray50") +
  geom_point(aes(colour = groups, shape = groups), size = 3.5 ) +
  geom_vline(aes(xintercept = 1), linetype = 2) +
  scale_shape_manual(values = rep(15, 5)) +
  scale_fill_manual(values = rep(c("#ffffff00", "#f0f0f090"), 9)[-1],
                    guide = "none") +
  scale_x_log10() +
  #coord_cartesian(xlim = c(0.01, 100)) +
  facet_grid(labels~., switch = "y") +
  theme_bw() +
  theme(panel.spacing.y = unit(0, "points"),
        panel.border = element_blank(),
        axis.text.y = element_blank(),
        plot.title = element_text(hjust = 0.5),
        axis.ticks.length.y = unit(0, "points"),
        strip.text.y.left = element_text(angle = 0),
        strip.background.y = element_blank(),
        strip.placement = "outside",
        axis.line = element_line()
  ) + 
  ylab("") + 
  xlab("Odds Ratio") + 
  ggtitle("Recent Exposure to Plasmodium falciparum") 

额外调整说明 #

如果需要组内按Odds Ratio值排序，只需将排序代码替换为：

# 组内按OR值从小到大排序
pfhist_df <- pfhist_df[order(pfhist_df$groups, pfhist_df$boxOdds), ]

# 若要从大到小排序，使用负号
pfhist_df <- pfhist_df[order(pfhist_df$groups, -pfhist_df$boxOdds), ]

内容的提问来源于stack exchange，提问作者bellbyrne