对多个变量使用 dplyr 和 ggplot

Question

我正在使用 dplyr 对我的数据进行子集化并为一个变量生成多个图表，但我想对多个变量使用相同的子集。 对于样本数据reprex ，

reprex <- structure(list(daycode = c("a", "a", "a", "a", "a", "a", "b", 
"b", "b", "b", "b", "b", "c", "c", "c", "c", "c", "c"), status = c("1", 
"2", "1", "2", "1", "2", "1", "2", "1", "2", "1", "2", "1", "2", 
"1", "2", "1", "2"), IHT = c(20, 65, 31, 62, 19, 58, 27, 27, 
28, 28, 23, 25, 27, 26, 59, 34, 16, 18), TJ = c(19L, 141L, 80L, 
181L, 24L, 166L, 65L, 103L, 40L, 98L, 45L, 88L, 59L, 86L, 54L, 
98L, 21L, 74L)), class = "data.frame", row.names = c(NA, -18L
))

并使用代码

d.plots <- reprex %>%
  group_by(status, daycode) %>%
  summarise(day.TJ = mean(TJ), sdtj = sd(TJ),
            day.IHT = mean(IHT), 
            sdIHT = sd(IHT)) %>%
  mutate(sdTJ.hi = day.TJ + sdtj, sdTJ.lo = day.TJ - sdtj, sdIHT.hi = day.IHT + sdIHT,
         sdIHT.lo = day.IHT - sdIHT) %>%
  do(plots = ggplot(data = .) +
       aes(x = daycode, y = day.TJ, group = 1) +
       geom_col(fill = "red")+
       geom_text(aes(y = day.TJ + 2, label = signif(day.TJ, 2))) +
       geom_line(aes(y = sdTJ.hi), size = 0.8) +
       geom_line(aes(y = sdTJ.lo), size = 0.8)+
       labs(x = "Day", y = "Avg Jumps" ) +
       theme_bw() +
       ggtitle(unique(.$status)))
d.plots$plots

我可以获得变量TJ的两个图表，但也想获得变量 IHT 的相同图表。

Answer 1

我会将 pivot 和 dataframe 转换为更长的格式，因此您将在列中拥有变量（ IHT或JT ），并且可以在不重复代码的情况下使用它。

我会这样做并使用您的代码来准备绘图。 我还将提出更好的数据可视化； 当然我不知道你打算用这些做什么，所以可以随意忽略它。

library(dplyr)
library(tidyr)
library(ggplot2)

d.plots<-reprex %>% 
  pivot_longer(cols = c('IHT', 'TJ')) %>%
  group_by(status, name, daycode) %>%
  summarise(day = mean(value),
            sd = sd(value)
  ) %>%
  mutate(sd.hi = day + sd, 
         sd.lo = day - sd) %>%
  do(plots = ggplot(data = .) +
       aes(x = daycode, y = day, group = 1) +
       geom_col(fill = "red")+
       geom_text(aes(y = day + 2, label = signif(day, 2))) +
       geom_line(aes(y = sd.hi), size = 0.8) +
       geom_line(aes(y = sd.lo), size = 0.8)+
       labs(x = "Day", y = "Avg Jumps" ) +
       theme_bw() +
       ggtitle(unique(paste(.$status, .$name))))
d.plots$plots
#> [[1]]

#> 
#> [[2]]

#> 
#> [[3]]

#> 
#> [[4]]

这是我提出的可视化，它用点交换条并利用facet特征：

reprex %>% 
  pivot_longer(cols = c('IHT', 'TJ')) %>%
  group_by(status, name, daycode) %>%
  summarise(day = mean(value),
            sd = sd(value)
  ) %>%
  mutate(sd.hi = day + sd, 
         sd.lo = day - sd) %>% 
  ggplot(aes(x = daycode, y = day, group = 1)) +
  facet_grid(status ~ name) +
  geom_point(color = 'red2') +
  geom_text(aes(label = signif(day, 2)), nudge_y = 10) +
  geom_ribbon(aes(ymin = sd.lo, ymax = sd.hi), alpha = .2) +
  theme_bw() 

^{由代表 package (v0.3.0) 于 2020 年 6 月 16 日创建}