ggplot：根据未显示的变量有条件地格式化点

Question

如果p得分大于0.1，我希望在下图中将该点显示为空洞，但我无法弄清楚如何使该点的格式以条件的值为条件p：

可以使用以下代码复制数据框：

structure(list(metric = structure(1:6, .Label = c("Viewers", 
"Views", "Chatters", "Chats", "Diamond Senders", "Diamonds"), class = "factor"), 
    group = c("Views", "Views", "Chats", "Chats", "Diamonds", 
    "Diamonds"), avg = c(-0.0264387458383021, -0.0515928340053015, 
    0.0097420053047867, 0.144967615821856, 0.347281593023006, 
    -0.25567948486292), err_low = c(-0.0664593483464582, -0.139178443719337, 
    -0.0595290104397667, -0.0490217193008488, -0.146892412146765, 
    -1.24468943017821), err_high = c(0.013581856669854, 0.0359927757087344, 
    0.0790130210493401, 0.33895695094456, 0.841455598192777, 
    0.733330460452365), p = c(0.277195141257042, 0.332587114951675, 
    0.817060323071668, 0.218997456845286, 0.247710490053607, 
    0.670666890004374), ord = c(1, 2, 3, 4, 5, 6)), .Names = c("metric", 
"group", "avg", "err_low", "err_high", "p", "ord"), row.names = c(NA, 
-6L), class = "data.frame")

图表可以使用ggplot2复制：

ggplot(datac, aes(x=metric, y=avg, color=group)) +geom_hline(yintercept = 0)+
      geom_label(aes(label=paste0(round(avg*100,2),"%")),nudge_x=-0.35, size=2.5) +geom_point(size=6)+ 
    geom_errorbar(position=position_dodge(.9), width=.25, aes(ymin=err_low, ymax=err_high)) +
     theme_minimal()+  scale_color_manual(values=c("Views"="blue", "Chats"="orange","Diamonds"="dark green")) +
     scale_y_continuous(labels = scales::percent,breaks=c(seq(-2,+2,by=0.2))) + labs(title="Change in Test vs. Control", 
         x="Experiment Metrics ", y = "% change")

感谢您的解决方案！

Answer 1

最简单的方法是使用所需的映射创建另一个因子列并将其应用于图表。我用了p>示例中的0.5

library(dplyr)
datac %>%
  mutate(p1= factor(ifelse(p > 0.5, 1, 0)))%>%
  ggplot(aes(x=metric, y=avg, color=group))+
  geom_hline(yintercept = 0)+
  geom_label(aes(label=paste0(round(avg*100,2),"%")),nudge_x=-0.35, size=2.5)+
  geom_point(size=6, aes(shape = p1))+
  geom_errorbar(position=position_dodge(.9), width=.25, aes(ymin=err_low, ymax=err_high))+
  theme_minimal()+
  scale_color_manual(values=c("Views"="blue", "Chats"="orange","Diamonds"="dark green")) +
  scale_y_continuous(labels = scales::percent,breaks=c(seq(-2,+2,by=0.2)))+
  labs(title="Change in Test vs. Control", x="Experiment Metrics ", y = "% change")+
  scale_shape_manual("p > 0.5", values=c("0" = 1, "1" = 16))