R - RStudio中每页多个图

时间：2016-07-30 04:34:16

标签： r plot rstudio

我有三个降水数据集：一个观察和两组模型：原始和偏差校正。每组包含五个模型：CanESM2, GFDL.ESM2M, MRI.CGCM3, NorESM1.M, inmcm4。

以下代码重现了我的数据：

obs <- c(8.12903225806452, 4.65483870967742, 4.00967741935484, 2.82903225806452, 
1.02258064516129, 4.94838709677419, 2.03548387096774, 4.56129032258065, 
6.66451612903226, 10.4129032258065, 3.7741935483871, 9.93870967741936)

uncorrected <- structure(list(CanESM2 = c(3.71099209785461, 6.68828105926514, 
4.2675461769104, 4.29092979431152, 6.47999143600464, 6.30958032608032, 
3.28256869316101, 1.06898427009583, 4.01592206954956, 7.88787031173706, 
10.8663206100464, 9.14756298065186), GFDL.ESM2M = c(4.56643295288086, 
8.36136913299561, 4.6934700012207, 4.62750291824341, 6.34791326522827, 
0.916408598423004, 0.111764870584011, 0.570569038391113, 3.05491662025452, 
3.29908037185669, 5.98263263702393, 7.52161026000977), MRI.CGCM3 = c(3.94400858879089, 
2.48861312866211, 0.663281202316284, 0.422377318143845, 1.49548053741455, 
1.8869035243988, 5.35201597213745, 3.94722390174866, 5.0041446685791, 
6.11909627914429, 6.37435054779053, 5.9401068687439), NorESM1.M = c(8.83542060852051, 
8.36209106445312, 3.22844076156616, 2.86131143569946, 2.05858564376831, 
2.39689040184021, 1.06175291538239, 1.03921580314636, 2.49208211898804, 
6.83605766296387, 7.95251750946045, 4.8852391242981), inmcm4 = c(5.88600730895996, 
8.03514385223389, 3.86212635040283, 1.51728367805481, 3.46821308135986, 
1.30993270874023, 0.497318655252457, 0.589299857616425, 1.63358855247498, 
3.61927103996277, 4.99840354919434, 10.5596942901611)), .Names = c("CanESM2", 
"GFDL.ESM2M", "MRI.CGCM3", "NorESM1.M", "inmcm4"), class = "data.frame", row.names = c(NA, 
-12L))

corrected <- structure(list(CanESM2 = c(4.69295692443848, 6.2362003326416, 
4.48602199554443, 4.21368360519409, 6.02497625350952, 5.94498348236084, 
3.92529225349426, 1.07059073448181, 3.92306709289551, 8.00215721130371, 
10.9688482284546, 9.95723056793213), GFDL.ESM2M = c(6.74847173690796, 
10.7781000137329, 6.89084196090698, 7.5780816078186, 8.96726703643799, 
3.57111501693726, 1.20403492450714, 2.76221370697021, 4.98387718200684, 
5.27927255630493, 8.87845706939697, 9.95864582061768), MRI.CGCM3 = c(5.45173072814941, 
3.37272930145264, 2.20000958442688, 1.12101686000824, 2.67246603965759, 
3.10278224945068, 6.02392053604126, 4.70185708999634, 5.86025190353394, 
6.41299057006836, 7.1956262588501, 7.59840393066406), NorESM1.M = c(9.49540519714355, 
9.88559532165527, 4.86537599563599, 4.56189870834351, 3.85974431037903, 
4.07706260681152, 2.55634951591492, 2.63215637207031, 4.21092081069946, 
7.66422271728516, 8.75070095062256, 5.95497798919678), inmcm4 = c(8.21576976776123, 
11.1972188949585, 6.08181619644165, 3.58428621292114, 5.543297290802, 
3.46666693687439, 1.88991332054138, 1.94812619686127, 3.71879911422729, 
5.70409107208252, 6.85269260406494, 11.8774194717407)), .Names = c("CanESM2", 
"GFDL.ESM2M", "MRI.CGCM3", "NorESM1.M", "inmcm4"), class = "data.frame", row.names = c(NA, 
-12L))

我需要做的是在模型之间根据观察创建qq图。因此，我将有10个图（5个模型×2套模型）。

然后，我需要将所有绘图组合在一个单独的页面中，在5x2矩阵中，其中行将是每个模型（CanESM2，GFDL.ESM2M等），列将代表每组模型：原始和偏差校正。

我快到了。这是我到目前为止所做的：

par(mfrow=c(5,2))

for (i in 1:length(names(uncorrected))) {

  # Plot uncorrected vs obs
  qqplot(as.vector(obs), uncorrected[,i], xlim=c(0,14), ylim=c(0,14),
         xlab=expression("Obs. precip." ~ (mm ~ day^{-1})),
         ylab=expression("Mod. precip." ~ (mm ~ day^{-1})))
  abline(0, 1)

  # Then plot corrected vs obs
  qqplot(as.vector(obs), corrected[,i], xlim=c(0,14), ylim=c(0,14),
         xlab=expression("Obs. precip." ~ (mm ~ day^{-1})),
         ylab=expression("Mod. precip." ~ (mm ~ day^{-1})))
  abline(0, 1)

}

但我在RStudio上遇到错误：Error in plot.new() : figure margins too large

我可以在独立的R上获得它，但最终的数字看起来非常糟糕，每个情节都看起来“拉长”。

如何将所有这些图形与RStudio中可接受的外观结合起来？

2 个答案:

答案 0 :(得分：1)

问题的原因是mar参数对于绘制窗口大小相对较大，如@shayaa＆amp; @Vandenman说。这是一组用于多点的par()参数。通过改变Rstudio的绘图窗口或输出选项的高度来解决ylab切割的问题。

par.old <- par(no.readonly=T)    # save old par

par( mfrow = c(5, 2), mar = c(3, 3, 0.5, 0.5), tcl = -0.3, mgp = c(1.7, 0.4, 0) )

for (i in 1:length(names(uncorrected))) {
  # Plot uncorrected vs obs
  qqplot(as.vector(obs), uncorrected[,i], xlim=c(0,14), ylim=c(0,14),
         xlab=expression("Obs. precip." ~ (mm ~ day^{-1})),
         ylab=expression("Mod. precip." ~ (mm ~ day^{-1})))
  abline(0, 1)
  # Then plot corrected vs obs
  qqplot(as.vector(obs), corrected[,i], xlim=c(0,14), ylim=c(0,14),
         xlab=expression("Obs. precip." ~ (mm ~ day^{-1})),
         ylab=expression("Mod. precip." ~ (mm ~ day^{-1})))
  abline(0, 1)
}

par(par.old)    # read old par  (not necessary)

如果xlab和ylab是绘图的共同点，则可以通过在外边距处仅编写一次xlab和ylab来节省空间。

par( mfrow = c(5, 2), mar = c(1.2, 1.2, 0.5, 0.5), tcl = -0.3, mgp = c(1.7, 0.4, 0), 
     oma = c(2.5, 2.5, 0, 0) )

for (i in 1:length(names(uncorrected))) {
  # Plot uncorrected vs obs
  qqplot(as.vector(obs), uncorrected[,i], xlim=c(0,14), ylim=c(0,14), ann=F)
  abline(0, 1)
  # Then plot corrected vs obs
  qqplot(as.vector(obs), corrected[,i], xlim=c(0,14), ylim=c(0,14), ann=F)
  abline(0, 1)
}

mtext(side=1, expression("Obs. precip." ~ (mm ~ day^{-1})), outer=T, line=1.1)   
mtext(side=2, expression("Mod. precip." ~ (mm ~ day^{-1})), outer=T, line=0.6)

enter image description here enter image description here

[EDITED]

除非你想要复杂的分割，否则我认为你不需要使用layout（）。

par(mfrow = c(3, 4),  mar = c(1.2, 1.2, 0.5, 0.5), tcl = -0.3, mgp = c(1.7, 0.4, 0), 
    oma = c(2.5, 2.5, 0, 0) )

for (i in 1:length(names(uncorrected))) {
  qqplot(as.vector(obs), uncorrected[,i], xlim=c(0,14), ylim=c(0,14), ann=F)
  abline(0, 1)
  legend("topleft", paste0("uncorrected",i), bty="n")  # example
  qqplot(as.vector(obs), corrected[,i], xlim=c(0,14), ylim=c(0,14), ann=F)
  abline(0, 1)
  legend("topleft", paste0("corrected", i), bty="n")   # example
}
for(i in 1:2) plot(0, type="n", ann=F, axes=F) # make two blank graphs

mtext(side=1, expression("Obs. precip." ~ (mm ~ day^{-1})), outer=T, line=1.1)
mtext(side=2, expression("Mod. precip." ~ (mm ~ day^{-1})), outer=T, line=0.6)

如果要使用layout（），最好不使用par()使用相同的mfrow()参数。 （下面的代码相当于上面的代码）

m <- matrix(c(1:10, 0, 0), byrow=T, ncol=4)
layout(m)

par( mar = c(1.2, 1.2, 0.5, 0.5), tcl = -0.3, mgp = c(1.7, 0.4, 0), 
     oma = c(2.5, 2.5, 0, 0) )

for (i in 1:length(names(uncorrected))) {
  qqplot(as.vector(obs), uncorrected[,i], xlim=c(0,14), ylim=c(0,14), ann=F)
  abline(0, 1)
  legend("topleft", paste0("uncorrected",i), bty="n")  # example
  qqplot(as.vector(obs), corrected[,i], xlim=c(0,14), ylim=c(0,14), ann=F)
  abline(0, 1)
  legend("topleft", paste0("corrected", i), bty="n")   # example
}

mtext(side=1, expression("Obs. precip." ~ (mm ~ day^{-1})), outer=T, line=1.1)   
mtext(side=2, expression("Mod. precip." ~ (mm ~ day^{-1})), outer=T, line=0.6)

如果你想重复NO.9＆amp; 10个图表，您应该更改for()参数（即for ( i in c(1:5, 5) )）（如果使用layout()，m <- matrix(c(1:12), byrow=T, ncol=4)。在layout()中，相同的数字表示对接）< / p>

enter image description here

答案 1 :(得分：1)

Error in plot.new() : figure margins too large可能是一个与计算机相关的错误。如果您只是将Rstudio中的plotwindow调整为尽可能大，然后重新运行代码，它是否仍会发生？如果是，可能Rstudio无法在您的机器上显示情节。您可以通过在绘图之前调用x11()来打开一个新的窗口（就像基地R那样），或者你可以使用类似的东西来解决这个问题：

pdf('test.pdf')
# make plots
dev.off()

创建包含图表的pdf文档。

相关问题
最新问题