r - 箭头函数不适用于 R 中的错​​误栏

Question

我非常感谢这个论坛中乐于助人的社区的帮助，解决我在使用 R 时遇到的一个小问题，以及在尝试绘制错误栏时的箭头功能。我选择使用箭头函数来生成我的错误栏，因为它们可以简单地添加到我已经创建的图表中。

我已经成功地使用 plot() 为下表绘制了两个散点图

     X sitting.time exercise.time se.sit.time se.ex.time
     1     535.7308      70.08000    45.39388   13.23958
     2     512.8800      65.80000    50.92198   13.98714
     3     500.2727      80.09091    47.00300   13.90255
     4     504.4762      77.35000    48.38485   17.22151
     5     487.0588      84.00000    51.62004   14.62120
     6     487.3333      83.53333    51.15082   17.52736
     7     462.5294      83.52941    49.97000   17.61185
     8     458.1111      71.78947    47.30922   12.64716
     9     471.9333      73.66667    57.13613   16.97271
    10     517.1333      84.92857    62.34817   18.61450
    11     526.5625      73.93750    50.25767   17.67990
    12     484.0000      96.22222    73.67854   23.97266

在表格中，sitting.time 和 exercise.time 是我的 y 值，X 是 x 值，se.sit.time 和se.ex.time 是标准错误。

我已经成功用于生成两个图(包括图例)的命令如下:

plot(weekly.times$X,weekly.times$sitting.time,
 xlim=c(1,12),ylim=c(0,600),
 xaxt="n", yaxt="n",xlab="Week", 
 ylab="Time (minutes)", type="p",col="red")
 axis(side=1,at=c(1:12))
 axis(side=2,at=seq(0,600,by=50))
 lines(weekly.times$X,weekly.times$sitting.time, 
   xlab="Week", ylab="Time", col="red")
 par(new=TRUE)
 plot(weekly.times$X,weekly.times$exercise.time, 
  xlab='', ylab='', axes=FALSE, col="blue")
 lines(weekly.times$X,weekly.times$exercise.time, 
   xlab='', ylab='', col="blue")
 legend("bottom",inset=c(-0.3,0), 
     legend=c("Sitting Time", "Exercise Time"), 
     col=c("red", "blue"), lty=1:2, cex=0.8)

但是，在尝试在第一个图之后添加 arrows() 之后，我没有看到任何错误栏以与主题类似帖子的原始答案相同的方式生成:Scatter plot with error bars

arrows(weekly.times$sitting.time, sitCI.down, 
  weekly.times$sitting.time, sitCI.up, length=0.05, angle=90, code=3)

第一个图的输出，我想在其中添加误差线。在此情节之后，我还运行了箭头功能:

我在图表中包含两个图的输出。我最终希望这两个图表都有误差线。我也试过在这里运行 arrows() 但没有成功:

有谁知道我可能做错了什么，以及为什么错误图没有显示在图形中？

---

在遵循@rbatt 的解决方案之后，我得到了可以处理错误图的图表，如下所示:

我使用了以下代码:

p1.ylim <- range(c(weekly.times$sitting.time, sitCI.down, sitCI.up))
p1.xlim <- range(weekly.times$X)
par(mar=c(4,4,0.5,4))
plot(weekly.times$X, weekly.times$sitting.time, xlim=p1.xlim, ylim=p1.ylim, ylab="Sitting Time (minutes)", xlab="Week", type="o",col="red", xaxt="n", yaxt="n")
axis(side=1,at=weekly.times$X)
axis(side=2,at=seq(p1.ylim[1],p1.ylim[2],by=50))
red2 <- adjustcolor("red", alpha.f=0.25)
arrows(weekly.times$X, sitCI.down, weekly.times$X, sitCI.up, length=0.05, angle=90, code=3, col=red2)
par(new=TRUE)
plot(weekly.times$X,weekly.times$exercise.time, xlab='', ylab='', axes=FALSE, col="blue", type="o")
axis(side=4)
blue2 <- adjustcolor("blue", alpha.f=0.25)
arrows(weekly.times$X, exCI.down, weekly.times$X, exCI.up, length=0.05, angle=90, code=3, col=blue2)
mtext("Exercise Time (minutes)", side=4, line=3)
legend("bottom",inset=c(-0.3,0), legend=c("Sitting Time", "Exercise Time"), col=c("red", "blue"), lty=1:2, cex=0.8)

但是，我现在有如下新数据，它没有得到适合的误差线。

week sitting.time exercise.time se.sit.time se.ex.time
1     1     3643.087      471.5833    349.5165   98.06667
2     2     3516.478      472.7917    337.8359   99.83149
3     3     3373.000      535.6667    344.9220  103.70350
4     4     3480.952      510.2857    352.2140  118.32880
5     5     3235.882      571.3529    345.0725  104.53620
6     6     3359.000      587.1333    349.3253  122.10290
7     7     3246.250      554.1250    358.0210  131.71030
8     8     3173.889      502.7895    326.9745   88.58391
9     9     3438.643      513.0667    354.2999  119.08770
10   10     3581.154      807.7143    323.0524  230.90890
11   11     3661.250      693.7059    357.1707  212.60630
12   12     3297.500      909.0000    589.6690  276.65540

使用相同的代码，我得到以下图:

你知道我在轴代码中哪里出错导致蓝线的错误栏不显示吗？

谢谢。

Answer 1

您遇到了两个主要问题。首先，您向 arrows 的 x0 和 x1 参数提供了一个 y 轴变量 weekly.times$sitting.time ();我已将它们更改为 weekly.times$X，即星期，位于您的 x 轴上。其次，您尝试在调用 par(new=TRUE) 并添加练习时间后添加箭头，练习时间的范围在 ~60 到 ~100 之间，因为 ylim 不是set 和你的 arrows 在 500-600 范围内，它们将被创建远离情节(即使第一个问题已经解决)。

以下代码对您的原始代码进行了最少的更改，以生成合适的图形。请注意，您没有提供对象 sitCI.down 和 sitCI.up 的数据，所以我只在坐着时间周围做了 ±10%。此外，我首先将 dput() 的输出包含在我从您的问题中手动复制的 data.frame 中；将来，如果您对数据使用 dput() 并将输出复制到您的问题中，那么重现结果会容易得多。

使用 dput 的输出创建数据集

weekly.times <- structure(list(X = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12), 
    sitting.time = c(535.7308, 512.88, 500.2727, 504.4762, 487.0588, 
    487.3333, 462.5294, 458.1111, 471.9333, 517.1333, 526.5625, 
    484), exercise.time = c(70.08, 65.8, 80.09091, 77.35, 84, 
    83.53333, 83.52941, 71.78947, 73.66667, 84.92857, 73.9375, 
    96.22222), se.sit.time = c(45.39388, 50.92198, 47.003, 48.38485, 
    51.62004, 51.15082, 49.97, 47.30922, 57.13613, 62.34817, 
    50.25767, 73.67854), se.ex.time = c(13.23958, 13.98714, 13.90255, 
    17.22151, 14.6212, 17.52736, 17.61185, 12.64716, 16.97271, 
    18.6145, 17.6799, 23.97266)), .Names = c("X", "sitting.time", 
"exercise.time", "se.sit.time", "se.ex.time"), row.names = c(NA, 
-12L), class = "data.frame")

在对原始代码进行最少更改的情况下生成所需的图

plot(weekly.times$X, weekly.times$sitting.time, xlim=c(1,12), ylim=c(0,600), ylab="Time (minutes)", type="p",col="red")
axis(side=1,at=c(1:12))
axis(side=2,at=seq(0,600,by=50))
lines(weekly.times$X, weekly.times$sitting.time, xlab="Week", ylab="Time", col="red")

sitCI.down <- weekly.times$sitting.time + c(-0.1)*weekly.times$sitting.time
sitCI.up <-  weekly.times$sitting.time + c(0.1)*weekly.times$sitting.time
arrows(weekly.times$X, sitCI.down, weekly.times$X, sitCI.up, length=0.05, angle=90, code=3)

par(new=TRUE)
plot(weekly.times$X,weekly.times$exercise.time, xlab='', ylab='', axes=FALSE, col="blue")
lines(weekly.times$X,weekly.times$exercise.time, xlab='', ylab='', col="blue")
legend("bottom",inset=c(-0.3,0), legend=c("Sitting Time", "Exercise Time"), col=c("red", "blue"), lty=1:2, cex=0.8)

所以上面的代码应该可以解决你的问题。但是，我想就您的身材提供一些额外的帮助/风格建议。请参阅下面的代码。

更简洁、更简洁的方法(我的意见)

# Calculate CI's first to account for them in ylim of plot
sitCI.down <- weekly.times$sitting.time + c(-0.1)*weekly.times$sitting.time
sitCI.up <-  weekly.times$sitting.time + c(0.1)*weekly.times$sitting.time

# I very frequently follow this idiom for making sure the ylims are set properly
# It is better to define ylims programmatically than by using magic numbers like c(0,600)
p1.ylim <- range(c(weekly.times$sitting.time, sitCI.down, sitCI.up))
p1.xlim <- range(weekly.times$X) # same idea for xlim, but simpler case

# When creating your plot, you can set type="o" do draw the points and the lines at the same time
# Also, if you're going to create the axes with axis(), you can also not draw them in the plot
# Turn axes off using xaxt="n" and yaxt="n"
# Finally, if you are going to use par(new=TRUE), you are creating an 
# independent y-axis, and you should leave room in the right margin to draw it
par(mar=c(4,4,0.5,4))
plot(weekly.times$X, weekly.times$sitting.time, xlim=p1.xlim, ylim=p1.ylim, ylab="Sitting Time (minutes)", xlab="Week", type="o",col="red", xaxt="n", yaxt="n") # I would also change the axis label

# When drawing axes, avoid magic numbers
axis(side=1,at=weekly.times$X)
axis(side=2,at=seq(p1.ylim[1],p1.ylim[2],by=50))

# When drawing the errors, you might want to keep the color of the line
# but maybe make it a bit lighter so it isn't easily confused
# making the error bars fainter makes the figure much easier on the eyes
red2 <- adjustcolor("red", alpha.f=0.25)
arrows(weekly.times$X, sitCI.down, weekly.times$X, sitCI.up, length=0.05, angle=90, code=3, col=red2)

par(new=TRUE)
# again in this plot, just use type="o" to accomplish in 1 line what you're doing in 2
plot(weekly.times$X,weekly.times$exercise.time, xlab='', ylab='', axes=FALSE, col="blue", type="o")
axis(side=4) # add the axis for the blue line
mtext("Exercise Time (minutes)", side=4, line=3)
legend("bottom",inset=c(-0.3,0), legend=c("Sitting Time", "Exercise Time"), col=c("red", "blue"), lty=1:2, cex=0.8)