如何根据 facet wrap 中的 2 组中的 1 组对条形图进行排序？

Question

希望有人能帮我解决以下问题：我想显示 2 个不同组 (gruppe) 的不同实验室参数 (parameter) 的值 (avg)。此外，我想根据 3 个不同方面随时间的变化（性能）绘制此信息。这是数据集的一小部分：

# A tibble: 402 x 4
# Groups:   gruppe, parameter [134]
   gruppe parameter                      performance     avg
   <chr>  <chr>                          <chr>         <dbl>
 1 DGE    ACPA(citrull. Prot.-Ak) EIA/Se change_t1t0 NaN    
 2 DGE    ACPA(citrull. Prot.-Ak) EIA/Se change_t2t0  37.6  
 3 DGE    ACPA(citrull. Prot.-Ak) EIA/Se change_t3t0 NaN    
 4 Fasten Apolipoprot. A1 HP             change_t1t0  41.2 
 5 DGE    Apolipoprot. A1 HP             change_t2t0 NaN    
 6 DGE    Apolipoprot. A1 HP             change_t3t0 NaN    
 7 DGE    Apolipoprotein B               change_t1t0 NaN    
 8 DGE    Apolipoprotein B               change_t2t0 NaN    
 9 Fasten Apolipoprotein B               change_t3t0 NaN    
10 DGE    aPTT Pathromtin SL             change_t1t0   0.571
# … with 392 more rows

使用这段代码完全没问题：

#Create labels for 3 facets
lab_labels <- c("Change from Baseline to Day 7 [%]",
                "Change from Baseline to Week 6 [%]",
                "Change from Baseline to Week 12 [%]")

names(lab_labels) <- c("change_t1t0",
                       "change_t2t0",
                       "change_t3t0")

labor_summ_long %>%
  filter(parameter %in% c("Hämatokrit (l/l)","Hämoglobin", "Leukozyten","MCV", "MCH", "MCHC", "RDW-CV", "Thromobzyten","MPV")) %>%
  arrange(desc(avg))%>%
  group_by(gruppe, performance)%>%
  ggplot(aes(x=reorder(parameter,avg), y=avg, group=gruppe, fill = gruppe))+
  geom_col(position = position_dodge())+
  facet_wrap(~performance, 
             scales ="free_y", 
             dir="v",
             labeller = labeller(performance = lab_labels))+
  ylab("") + 
  xlab("") + 
  labs(color="", linetype="")+
  theme_pubclean()+
  theme(strip.background=element_rect(fill="lightgrey"),
        strip.text = element_text(face="bold"),
        legend.position = "bottom",
        legend.title=element_blank())+
  theme(axis.text.x = element_text(angle=45, hjust=1, vjust = 1))+
  scale_x_discrete(labels = c("Hämoglobin"="Hemoglobin", "Leukozyten" = "Leucocytes",
                              "MCV", "MCH", "MCHC", "RDW-CV", "Thromobzyten"="Thrombocytes",
                              "MPV", "Hämatokrit (l/l)"="Hematocrite"))+
  scale_fill_discrete(labels=c('DGE', "Fasten"='Fasting'))

This is how the plot looks like

我遗漏了什么并且未能找到解决方案：我想订购酒吧...

按照从高到低的avg-value
禁食组（蓝色条）
从基线到第 7 天的表现（change_t1t0），也就是第一个方面。

我用 arrange、sort 等来欺骗，但无法将上述所有条件放在一起。

你有什么想法吗？非常感谢！

Answer 1

问题是 reorder 通过取每个 parameter 的所有值的平均值重新排序，而不考虑任何分组。

根据您的情况调整答案并使用一些随机示例数据来模拟您的真实数据，这可以像这样实现：

辅助函数 reorder_where 允许根据附加条件对类别进行排序，例如在你的情况下 gruppe == "Fasten" & performance == "change_t1t0" 是 TRUE

library(dplyr)
library(ggplot2)

reorder_where <- function (x, by, where, fun = mean, ...) {
  xx <- x[where]
  byby <- by[where]
  byby <- tapply(byby, xx, FUN = fun, ...)[x]
  reorder(x, byby)
}

labor_summ_long %>%
  filter(parameter %in% c("Hämatokrit (l/l)","Hämoglobin", "Leukozyten","MCV", "MCH", "MCHC", "RDW-CV", "Thromobzyten","MPV")) %>%
  ggplot(aes(x=reorder_where(parameter, -avg, gruppe == "Fasten" & performance == "change_t1t0"), y=avg, group=gruppe, fill = gruppe))+
  geom_col(position = position_dodge())+
  facet_wrap(~performance, 
             scales ="free_y", 
             dir="v",
             labeller = labeller(performance = lab_labels))+
  ylab("") + 
  xlab("") + 
  labs(color="", linetype="")+
  #theme_pubclean()+
  theme(strip.background=element_rect(fill="lightgrey"),
        strip.text = element_text(face="bold"),
        legend.position = "bottom",
        legend.title=element_blank())+
  theme(axis.text.x = element_text(angle=45, hjust=1, vjust = 1))+
  scale_x_discrete(labels = c("Hämoglobin"="Hemoglobin", "Leukozyten" = "Leucocytes",
                              "MCV", "MCH", "MCHC", "RDW-CV", "Thromobzyten"="Thrombocytes",
                              "MPV", "Hämatokrit (l/l)"="Hematocrite"))+
  scale_fill_discrete(labels=c('DGE', "Fasten"='Fasting'))

数据

set.seed(42)

labor_summ_long <- data.frame(
  parameter = sample(c("Hämatokrit (l/l)","Hämoglobin", "Leukozyten","MCV", "MCH", "MCHC", "RDW-CV", "Thromobzyten","MPV"), 100, replace = TRUE),
  gruppe = sample(c("DGE", "Fasten"), 100, replace = TRUE),
  performance = sample(c("change_t1t0",
                         "change_t2t0",
                         "change_t3t0"), 100, replace = TRUE),
  avg = runif(100, 0, 50)
)
labor_summ_long <- dplyr::distinct(labor_summ_long, parameter, gruppe, performance, .keep_all = TRUE)

如何根据 facet wrap 中的 2 组中的 1 组对条形图进行排序？

How to sort bars according to 1 of 2 groups in a facet wrap?

sorting

r

bar-chart

ggplot2

facet-wrap