如何为双轴设置不同的范围并分配适当的标准误差

Question

我使用双 y 轴绘制条形图和折线图。我想修改单个y轴的范围，例如折线图在条形图上方。在这种情况下，我希望Does 的范围为 1-1.5，Wight 的范围为 0.20-0.5，因此线条会稍微悬停在条形图上。但是，当我在sec.axis 之后设置limits 时，仅适用于辅助y 轴并且条形图消失了。我尝试添加break 来给出一个范围，但它也没有成功。我想知道是否可以单独操纵两个 y 轴？

另外，当应用双轴时，是否可以为每个图（条形图一个，折线图一个）设置标准误差？我应用了两个 stat_summary 以获得 2 个不同的 mean_se（标准误差线），但是，只绘制了一个。任何建议都非常感谢！

# dummy data
burger<- tibble(
  Day = rep(1:4,2),
  Dose = c(0.5,0.6,0.4,0.3,0.5,0.6,0.4,0.3),
  Wight = c(0.95,0.92,0.93,0.98,0.95,0.92,0.93,0.98))


  # plot
  ggplot(data= burger, aes(x = Day, y = Dose),group = 1)+
    geom_bar(data= burger, aes(x = Day, y = Wight* (0.98/0.6)), stat = "identity") +
    stat_summary(fun.data = mean_se,geom= "errorbar",width=0.6,size=1,aes(y= burger$Dose, width=0.05))+
    stat_summary(fun.data = mean_se,geom= "errorbar",width=0.6,size=1,aes(y= burger$Wight, width=0.05))+
    geom_line (size = 2,aes(group=1))+
    geom_point(aes(colour=Day))  +
    scale_y_continuous(name = "Dose", 
                       sec.axis = sec_axis(trans =  ~.*(0.6/0.98) , name = "Wight",limits=c(0.25,1)))

Answer 1

在ggplot2 中，您控制主要的 y 轴，并转换次要的 y 轴。 plotly 允许您分别控制两个轴。

library(plotly)
plot_ly(data = burger) %>% 
  add_lines(x = ~Day, y = ~Dose,  name = "Dose") %>% 
  add_bars(x = ~Day, y = ~Wight, yaxis = "y2", name = "Wight") %>%
  layout(
    yaxis = list(showline = TRUE, side = "left", 
                 title = "Dose", range = c(0.2, 0.6)), 
    yaxis2 = list(showline = TRUE, side = "right", 
                  overlaying = "y",  title = "Wight", range = c(0.9, 1)), 
    showlegend = FALSE,  
    margin = list(pad = 0, b = 50, l = 50, r = 50))

Answer 2

我添加了一个 +1 和一个 -1 来将条形设置在同一行的高度，同样使用 coord_cartesian 您可以在不裁剪数据的情况下放大图表。 （如果您使用 ylim 代替，条形图会像发生在您身上一样消失）。 我也稍微调整了你的代码。

require(dplyr)
require(ggplot2)

# dummy data
burger <- tibble(Day   = 1:4,
                 Dose  = c(0.5, 0.6, 0.4, 0.3),
                 Wight = c(0.95, 0.92, 0.93, 0.98))


# plot
ggplot(data = burger, aes(x = Day, y = Dose))+
  geom_bar(aes(y = Wight * (0.98/0.6) - 1), stat = "identity") +
  geom_line(size = 2) +
  geom_point(aes(colour = Day))  +
  scale_y_continuous(name = "Dose", 
                     sec.axis = sec_axis(trans =  ~(.+1)*(0.6/0.98),
                                         name = "Wight")) +
  coord_cartesian(ylim = c(0.3, 0.6))

编辑：

此图表回答了您问题的第二部分。

首先，我创建了一个随机数据样本。

图表的所有对象都是用stat_summary 创建的。使用stat_summary，我可以将汇总函数应用于我的数据：意味着fun.y，当我需要从我的Y var 和mean_se 和fun.data 中获取单个结果时，当我需要返回一个可供geom errorbar 使用的向量时。

我添加了一些颜色来帮助您识别每件作品并将每个值与其错误栏直观地关联起来，但您可以设置最喜欢的颜色。

require(dplyr)
require(ggplot2)

set.seed(20)

# dummy data
burger<- tibble(
  Day   = rep(1:4, each = 10),
  Dose  = rnorm(40, 0.5, 0.1),
  Wight = rnorm(40, 0.9, 0.1))


# plot
ggplot(data = burger, aes(x = Day))+

  # bars and error
  stat_summary(aes(y = Wight * (0.98/0.6) - 1), 
               fun.y = mean, geom = "bar", 
               fill = "steelblue", alpha = 0.6) +
  stat_summary(aes(y = Wight * (0.98/0.6) - 1), 
               fun.data = mean_se, geom = "errorbar",
               colour = "steelblue", size = 1.5, width = 0.5) +

  # line/point and error
  stat_summary(aes(y = Dose), fun.data = mean_se, 
               geom = "errorbar", size = 1.5, width = 0.5,
               colour = "coral") +
  stat_summary(aes(y = Dose), fun.y = mean, geom = "line", 
               size = 0.5, colour = "coral") +
  stat_summary(aes(y = Dose, colour = Day), fun.y = mean, geom = "point", size = 3) +

  # prettier
  scale_y_continuous(name = "Dose", 
                     sec.axis = sec_axis(trans =  ~(.+1)*(0.6/0.98),
                                         name = "Wight")) +
  coord_cartesian(ylim = c(0.3, 0.6)) +
  theme_light() +
  scale_color_gradient2(mid = "black", high = "#FFBE5A")