ggplot分面图表各分面x轴因子自定义排序实现问题

需要绘制分面散点图，要求每个分面（按Pal分组）内的x轴（Effect）按数据中Value的递增顺序排列，同时保留分面独立背景色、点样式等设置。尝试用facetscales包的facet_grid_sc设置独立x轴刻度后，图表显示错乱。

初始数据集 #

dat<-data.frame(Pal=rep(c("A","B","C","D"),each=5),
            Rol=c("aa","aa","aa","aa","bb","aa","aa","aa","aa","aa","cc","cc","cc","cc","cc","aa","aa","aa","aa","aa"),
            Cel=rep(c("home","tree","hat","ball","pen","rope"),times=c(5,3,2,5,2,3)),
            Value=c(7701.1,59897.3,59897.3,59897.3,744438.1,1226.4,1454.6,1454.6,1454.6,1454.6,56600,92400,5010000,7010000,15740000,28.5,34.2,39.9,48.5,57),
            Col=c("black","red","black","black","red","red","red","black","black","black","red","red","red","red","red","red","black","black","black","black"),
            Effect=c("length","length","length","length","height","weight","length","length","age of youngest individual found miles from the closest coastline","age of youngest individual found miles from the closest coastline","pressure","speed","rate","rate","length","length","rate","rate","O2","fecundity")
            )

初始绘图代码（未实现排序） #

highlights<-data.frame(Pal=c("A","B","C","D"))
highlights$Pal<-factor(highlights$Pal,levels=c("A","B","C","D"))

ggplot() + 
  geom_rect(data=highlights,aes(xmin=-Inf, xmax=Inf, ymin=1, ymax=1000000000), fill=c("yellow","blue","red","green"), alpha=0.05) +
  geom_point(data = dat, aes(x=Effect, y=Value, shape=Rol, col=Col)) +
  scale_color_manual(breaks=unique(dat$Col), values=as.character(unique(dat$Col))) +
  labs(x="",y="Activity") + facet_grid(.~Pal, scales = "free_x")+ 
  scale_y_log10(limits=c(1,1000000000),breaks = c(1,10,100,1000,10000,100000,1000000,10000000,100000000,100000000,1000000000)) +
  scale_x_discrete(labels = function(x) str_wrap(x, width = 40)) +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
        panel.background = element_rect(fill = "white", colour = "white"),
        strip.background = element_rect(fill = "white", colour = "black"),
        legend.key = element_rect(fill = "white")) 

尝试的排序方法（导致图表错乱） #

用facetscales包设置分面独立x轴刻度，按每个Pal分组的Effect出现顺序排序，但图表显示异常：

my_scales <- list(
  "A" = scale_x_discrete(limits = as.character(unique(dat[which(dat$Pal=="A"),"Effect"]))),
  "B" = scale_x_discrete(limits = as.character(unique(dat[which(dat$Pal=="B"),"Effect"]))),
  "C" = scale_x_discrete(limits = as.character(unique(dat[which(dat$Pal=="C"),"Effect"]))),
  "D" = scale_x_discrete(limits = as.character(unique(dat[which(dat$Pal=="D"),"Effect"])))
)


ggplot() + 
  geom_rect(data=highlights,aes(xmin=-Inf, xmax=Inf, ymin=1, ymax=1000000000), fill=c("yellow","blue","red","green"), alpha=0.05) +
  geom_point(data = dat, aes(x=Effect, y=Value, shape=Rol, col=Col)) +
  scale_color_manual(breaks=unique(dat$Col), values=as.character(unique(dat$Col))) +
  labs(x="",y="Activity") + facet_grid_sc(cols = vars(Pal), scales=list(x = my_scales)) +
  scale_y_log10(limits=c(1,1000000000),breaks = c(1,10,100,1000,10000,100000,1000000,10000000,100000000,100000000,1000000000)) +
  scale_x_discrete(labels = function(x) str_wrap(x, width = 40)) +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
        panel.background = element_rect(fill = "white", colour = "white"),
        strip.background = element_rect(fill = "white", colour = "black"),
        legend.key = element_rect(fill = "white"))

补充测试数据集（更大规模，原方案失效） #

dat<-data.frame(Pal=rep(c("A","B","C","D"),times=c(20,40,60,30)),
                Rol=c(rep("aa",15),rep("bb",5),rep("aa",40),rep("cc",60),rep("aa",30)),
                Cel=rep(c("home","tree","hat","ball","arm","leg","beer","stick","pen","rope"),times=c(15,13,12,12,13,12,15,18,17,23)),
                Value=c(runif(n=20,min = 7000, max = 100000),runif(n=40,min = 100, max = 100000),runif(n=60,min = 1000, max = 1000000),runif(n=30,min = 100000, max = 10000000)),
                Col=rep(c("red","black"),each=5,times=15),
                Effect=c(rep(c("length","height"),times=c(15,5)),rep(c("weight","length","age of youngest individual found miles from the closest coastline"),times=c(10,7,23)),rep(c("pressure","speed","rate","length"),times=c(10,3,7,40)),rep(c("length","rate","O2","fecundity"),times=c(3,4,7,16)))
)

问题根源在于facetscales的刻度设置与全局scale_x_discrete冲突，且直接用unique(Effect)无法保证按Value递增排序。正确做法是为每个Pal分组内的Effect创建基于Value排序的自定义因子水平，再用常规facet_grid(scales="free_x")即可实现需求，无需依赖facetscales。

步骤1：为每个Pal分组的Effect设置排序后的因子水平 #

按Value均值排序（适合同一Effect对应多个Value的场景） #

library(dplyr)
library(ggplot2)
library(stringr)

dat_sorted <- dat %>%
  group_by(Pal, Effect) %>%
  summarise(mean_value = mean(Value), .groups = "drop") %>%
  arrange(Pal, mean_value) %>%
  group_by(Pal) %>%
  mutate(Effect_ordered = factor(Effect, levels = Effect)) %>%
  select(Pal, Effect, Effect_ordered) %>%
  left_join(dat, by = c("Pal", "Effect"))

按原始Value顺序排序（严格遵循数据中Value的递增顺序） #

dat_sorted <- dat %>%
  arrange(Pal, Value) %>%
  group_by(Pal) %>%
  mutate(Effect_ordered = factor(Effect, levels = unique(Effect))) %>%
  ungroup()

步骤2：绘制图表（用排序后的Effect_ordered作为x轴） #

highlights<-data.frame(Pal=c("A","B","C","D"))
highlights$Pal<-factor(highlights$Pal,levels=c("A","B","C","D"))

ggplot() + 
  # 分面背景色
  geom_rect(data=highlights,aes(xmin=-Inf, xmax=Inf, ymin=1, ymax=1e9), 
            fill=c("yellow","blue","red","green"), alpha=0.05) +
  # 散点图使用排序后的x轴
  geom_point(data = dat_sorted, aes(x=Effect_ordered, y=Value, shape=Rol, col=Col)) +
  # 颜色映射保持原设置
  scale_color_manual(breaks=unique(dat$Col), values=as.character(unique(dat$Col))) +
  labs(x="",y="Activity") + 
  # 分面设置自由x轴
  facet_grid(.~Pal, scales = "free_x")+ 
  # y轴对数刻度保持原样式
  scale_y_log10(limits=c(1,1e9),
                breaks = c(1,10,100,1000,10000,1e5,1e6,1e7,1e8,1e9)) +
  # x轴标签换行处理
  scale_x_discrete(labels = function(x) str_wrap(x, width = 40)) +
  # 主题设置保持原样式
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust=1),
        panel.background = element_rect(fill = "white", colour = "white"),
        strip.background = element_rect(fill = "white", colour = "black"),
        legend.key = element_rect(fill = "white"))

说明 #

• 此方法避免了facetscales的兼容性问题，同时保证每个分面的x轴按需求排序，图表显示正常。
• 两种排序方式可根据实际需求选择：若同一Effect对应多个Value，用均值排序更合理；若需严格遵循数据中Value的出现顺序，选择原始顺序排序。

内容的提问来源于stack exchange，提问作者Bithorax