如何在facet_wrap分组内用geom_path连接偏移(dodged)的分组点

问题核心是x轴变量选择和分组逻辑不匹配：你当前把Treatment设为x轴，每个Treatment对应的x位置上有两个Year的点，geom_path按Treatment分组时，只能在同一x位置连接两点，所以出现垂直路径。下面提供两种可行方案：

方案1：将Year设为x轴（推荐） #

把x轴换成Year，让同一Treatment的两个时间点处于不同x位置，再通过position_dodge让同一Year下的不同Treatment点偏移，这样路径就能正确连接同一Treatment的时间轨迹：

library(ggplot2)
library(stringr)

set.seed(123) # 固定随机数保证结果可复现
df <- data.frame (Treatment  = c("Drill_Herb", "Drill_Herb", "Drill_None", "Drill_None", "Herb_None","Herb_None", "Herb_Drill", "Herb_Drill"),
                  Comparison_Combo = c("Drill_Herb_TO_Drill_None", "Drill_Herb_TO_Drill_None", "Herb_None_TO_Herb_Drill", "Herb_None_TO_Herb_Drill"),
                  Ratio = sample(1:8, 8, replace=FALSE),
                  Year = c("2017", "2021","2017", "2021","2017", "2021","2017", "2021"))

ggplot(df, aes(x=Year, y=Ratio, shape=Treatment, color=Treatment)) +
  facet_wrap(~Comparison_Combo, drop = TRUE, scales = "free_x", nrow = 1) +
  geom_point(size=3.5, position=position_dodge(width=0.55))+
  geom_path(aes(group=Treatment), lwd=1, position=position_dodge(width=0.55))

关键调整： #

• x轴改为Year，让每个时间点对应独立的x位置
• shape/color映射到Treatment，方便区分不同处理组
• geom_path保持group=Treatment，配合position_dodge让路径和偏移后的点位置对应

方案2：手动计算x轴偏移量（保留Treatment为x轴） #

如果必须把Treatment作为x轴，可手动给不同Year的点添加x偏移，再按Treatment分组绘制路径：

library(ggplot2)

set.seed(123)
df <- data.frame (Treatment  = c("Drill_Herb", "Drill_Herb", "Drill_None", "Drill_None", "Herb_None","Herb_None", "Herb_Drill", "Herb_Drill"),
                  Comparison_Combo = c("Drill_Herb_TO_Drill_None", "Drill_Herb_TO_Drill_None", "Herb_None_TO_Herb_Drill", "Herb_None_TO_Herb_Drill"),
                  Ratio = sample(1:8, 8, replace=FALSE),
                  Year = c("2017", "2021","2017", "2021","2017", "2021","2017", "2021"))

# 手动计算偏移量：0.275是position_dodge(width=0.55)的一半
df$x_pos <- as.numeric(factor(df$Treatment)) + 
  ifelse(df$Year == "2017", -0.275, 0.275)

ggplot(df, aes(x=x_pos, y=Ratio, shape=Year)) +
  facet_wrap(~Comparison_Combo, drop = TRUE, scales = "free_x", nrow = 1) +
  geom_point(size=3.5)+
  geom_path(aes(group=Treatment), lwd=1) +
  # 替换x轴标签为Treatment名称
  scale_x_continuous(breaks=unique(as.numeric(factor(df$Treatment))), 
                     labels=unique(df$Treatment))

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