使用ggplot2中的geom_errorbar（）重新格式化数据以绘制校准曲线

简介：

## Here's what my data frame looks like. 
## I know it's ugly, but if you copy and paste it into your console it should work!
df_cal <- structure(list(ref.co.mean = c(1.23638284617457, 1.46466241535712, 
2.16020882959014, 2.55054760052641, 3.13141175081258, 3.86968879644661, 
6.5914211520901), ref.co.sd = c(0.0196205483139859, 0.0229279198586359, 
0.0172965018302434, 0.0164690175286326, 0.00583116470707786, 
0.00975072766851073, 0.0388826652553337), ref.co.se = c(0.00346845569085442, 
0.00193776290206006, 0.00166435666462165, 0.00127061228762621, 
0.000583116470707786, 0.00229826855196908, 0.00614788918523735
), ref.co.ci = c(0.00707396201972773, 0.00383130164529687, 
0.00329939297398704, 
0.0025085329371034, 0.00115702958592763, 0.00484892279298878,  
0.0124352796323718), id = c("48c", "48c", "48c", "48c", "48c", 
"48c", "48c"), aa34.co.mean = c(0, 0.248857142857143, 0.823777777777778, 
1.256, 1.886, 2.446, 4.54), aa34.co.sd = c(0, 0.0716567783084826, 
0.0660714166547489, 0.0777970497665622, 0.0518459255872629, 0, 
0.0690217357069497), aa34.co.se = c(0, 0.00605610310675521, 
0.0063577250318932, 0.00600217269807407, 0.00518459255872628, 0, 
0.0109132946446067), aa34.co.ci = c(0, 0.0119739921598931, 
0.0126034483753748, 0.0118499152368743, 0.0102873564420935, 0, 
0.0220742219853317), id = c("aa34", "aa34", "aa34", "aa34", "aa34", "aa34", 
"aa34"), aa35.co.mean = c(0.2915625, 0.801035714285714, 1.39911111111111, 
1.80436904761905, 2.45672, 3.02355555555556, 5.134975), aa35.co.sd = 
c(0.0691998633940125, 0.0474980316455754, 0.0846624379229758, 
0.0822798331713915, 0.0595577165445419, 
0.0178768075145867, 0.0243007072942329), aa35.co.se = c(0.0122329231657723, 
0.00401431635364878, 0.00814664688751334, 0.00634802694633388, 
0.00595577165445419, 0.00421360393984362, 0.00384227919014218), aa35.co.ci = 
c(0.0249492112853266, 0.00793701687349159, 0.0161497773125, 
0.0125327252345785, 0.0118175430765459, 0.00888992723110191, 
0.00777174323014678), id = c("aa35", "aa35", "aa35", "aa35", 
"aa35", "aa35", "aa35")), .Names = c("ref.co.mean", "ref.co.sd", 
"ref.co.se", "ref.co.ci", "id", "aa34.co.mean", "aa34.co.sd", 
"aa34.co.se", "aa34.co.ci", "id", "aa35.co.mean", "aa35.co.sd", 
"aa35.co.se", "aa35.co.ci", "id"), row.names = c(1L, 33L, 173L, 
281L, 449L, 549L, 567L), class = "data.frame")

## This code only gets half of the job done...
## 95% Confidence Intervals for Error Bars:
p <- ggplot(df_cal, aes(x=ref.co.mean, y=aa34.co.mean)) + 
    theme_bw() + 
    geom_errorbar(aes(ymin=aa34.co.mean-aa34.co.ci, 
                      ymax=aa34.co.mean+aa34.co.ci), width =.05) +
    xlab("Reference CO (ppm)") + 
    ylab("AA34 CO (ppm)") + 
    geom_smooth(method='lm', formula = y~x, se = FALSE) + 
    geom_point(size=2, shape = 21, fill="White") + 
    geom_abline(intercept = 0, slope = 1, color, linetype=2, color = "firebrick") + 
    ggtitle("CO Calibration @ 0% RH") + 
    theme(plot.title = element_text(hjust = 0.5)) + 
    annotate("rect", xmin = 4.80, xmax = 5.70, ymin = 0.70, ymax = 1.70, 
       fill="white", colour="red") +
    annotate("text", x=5.25, y=1.50, label= "R^2 == 0.994", parse=T) + 
    annotate("text", x=5.25, y=1.20, label= "alpha == -0.9490", parse=T) + 
    annotate("text", x=5.25, y=0.90, label= "beta == 0.849", parse=T)
p

我有三种不同空气质量测量的汇总统计数据框架。仪器名称为

aa34

、

aa35

和

48c

。他们每个人都以ppm为单位测量一氧化碳。我有宽格式的数据，其中每个向量是三种仪器的平均值、标准偏差、标准误差或95%置信区间

我想使用

ggplot（）

和

geom\u errorbar（）

绘制这些汇总统计数据，但我在将数据转换为长格式以及在

ggplot（）

中为颜色映射提供ID变量方面遇到了一些问题。我正在学习教程。下面是我想复制的图（当然用有毒烟雾替换豚鼠牙齿数据）。我一直在尝试添加一个额外的y变量，并让它们由

ID

变量进行颜色协调。我想要的输出将用三个

id

向量中的两个替换示例中的

supp

向量，即包含

aa34

和

aa35

的向量。我与

剂量

向量的当量为

ref.co.mean

，即我们的

x

变量。与

len

向量等价的是长格式的向量

aa34.co.mean

和

aa35.co.mean

数据：

## Here's what my data frame looks like. 
## I know it's ugly, but if you copy and paste it into your console it should work!
df_cal <- structure(list(ref.co.mean = c(1.23638284617457, 1.46466241535712, 
2.16020882959014, 2.55054760052641, 3.13141175081258, 3.86968879644661, 
6.5914211520901), ref.co.sd = c(0.0196205483139859, 0.0229279198586359, 
0.0172965018302434, 0.0164690175286326, 0.00583116470707786, 
0.00975072766851073, 0.0388826652553337), ref.co.se = c(0.00346845569085442, 
0.00193776290206006, 0.00166435666462165, 0.00127061228762621, 
0.000583116470707786, 0.00229826855196908, 0.00614788918523735
), ref.co.ci = c(0.00707396201972773, 0.00383130164529687, 
0.00329939297398704, 
0.0025085329371034, 0.00115702958592763, 0.00484892279298878,  
0.0124352796323718), id = c("48c", "48c", "48c", "48c", "48c", 
"48c", "48c"), aa34.co.mean = c(0, 0.248857142857143, 0.823777777777778, 
1.256, 1.886, 2.446, 4.54), aa34.co.sd = c(0, 0.0716567783084826, 
0.0660714166547489, 0.0777970497665622, 0.0518459255872629, 0, 
0.0690217357069497), aa34.co.se = c(0, 0.00605610310675521, 
0.0063577250318932, 0.00600217269807407, 0.00518459255872628, 0, 
0.0109132946446067), aa34.co.ci = c(0, 0.0119739921598931, 
0.0126034483753748, 0.0118499152368743, 0.0102873564420935, 0, 
0.0220742219853317), id = c("aa34", "aa34", "aa34", "aa34", "aa34", "aa34", 
"aa34"), aa35.co.mean = c(0.2915625, 0.801035714285714, 1.39911111111111, 
1.80436904761905, 2.45672, 3.02355555555556, 5.134975), aa35.co.sd = 
c(0.0691998633940125, 0.0474980316455754, 0.0846624379229758, 
0.0822798331713915, 0.0595577165445419, 
0.0178768075145867, 0.0243007072942329), aa35.co.se = c(0.0122329231657723, 
0.00401431635364878, 0.00814664688751334, 0.00634802694633388, 
0.00595577165445419, 0.00421360393984362, 0.00384227919014218), aa35.co.ci = 
c(0.0249492112853266, 0.00793701687349159, 0.0161497773125, 
0.0125327252345785, 0.0118175430765459, 0.00888992723110191, 
0.00777174323014678), id = c("aa35", "aa35", "aa35", "aa35", 
"aa35", "aa35", "aa35")), .Names = c("ref.co.mean", "ref.co.sd", 
"ref.co.se", "ref.co.ci", "id", "aa34.co.mean", "aa34.co.sd", 
"aa34.co.se", "aa34.co.ci", "id", "aa35.co.mean", "aa35.co.sd", 
"aa35.co.se", "aa35.co.ci", "id"), row.names = c(1L, 33L, 173L, 
281L, 449L, 549L, 567L), class = "data.frame")

## This code only gets half of the job done...
## 95% Confidence Intervals for Error Bars:
p <- ggplot(df_cal, aes(x=ref.co.mean, y=aa34.co.mean)) + 
    theme_bw() + 
    geom_errorbar(aes(ymin=aa34.co.mean-aa34.co.ci, 
                      ymax=aa34.co.mean+aa34.co.ci), width =.05) +
    xlab("Reference CO (ppm)") + 
    ylab("AA34 CO (ppm)") + 
    geom_smooth(method='lm', formula = y~x, se = FALSE) + 
    geom_point(size=2, shape = 21, fill="White") + 
    geom_abline(intercept = 0, slope = 1, color, linetype=2, color = "firebrick") + 
    ggtitle("CO Calibration @ 0% RH") + 
    theme(plot.title = element_text(hjust = 0.5)) + 
    annotate("rect", xmin = 4.80, xmax = 5.70, ymin = 0.70, ymax = 1.70, 
       fill="white", colour="red") +
    annotate("text", x=5.25, y=1.50, label= "R^2 == 0.994", parse=T) + 
    annotate("text", x=5.25, y=1.20, label= "alpha == -0.9490", parse=T) + 
    annotate("text", x=5.25, y=0.90, label= "beta == 0.849", parse=T)
p

##以下是我的数据帧的外观。
##我知道这很难看，但是如果你把它复制粘贴到你的控制台上，它应该会工作的！
df_cal这里切换到长格式的问题是，x轴有一个长度7的变量，y轴有两个组合长度14的变量。因此，此解决方案绑定行，以便两次包含引用（x轴）数据。然后在
ggplot
美学中使用
颜色
和
组

library(ggplot2)

df_aa34_2<-df_cal[,c(1:4,6:10)]#select first 'aa' group including reference data (48c)
df_aa35_2<-df_cal[,c(1:4,11:15)]#select second 'aa' group including reference data (48c)
names(df_aa34_2)<-names(df_aa35_2)#colnames must be the same for rbind function
DF<-rbind(df_aa34_2,df_aa35_2)#bind rows

p <- ggplot(DF,aes(x=ref.co.mean,y=aa35.co.mean,colour=id,group=id)) + 
  geom_errorbar(aes(ymin=aa35.co.mean-aa35.co.ci, 
                    ymax=aa35.co.mean+aa35.co.ci), width =.5) +
  xlab("Reference CO (ppm)") + 
  ylab("AA34 & 35 CO (ppm)") + 
  geom_smooth(method='lm', formula = y~x, se = FALSE) + 
  geom_point(size=2, shape = 21, fill="White") + 
  geom_abline(intercept = 0, slope = 1, color, linetype=2, color = "firebrick") + 
  ggtitle("CO Calibration @ 0% RH") + 
  theme(plot.title = element_text(hjust = 0.5)) + 
  annotate("rect", xmin = 4.80, xmax = 5.70, ymin = 0.70, ymax = 1.70, 
           fill="white", colour="red") +
  annotate("text", x=5.25, y=1.50, label= "R^2 == 0.994", parse=T) + 
  annotate("text", x=5.25, y=1.20, label= "alpha == -0.9490", parse=T) + 
  annotate("text", x=5.25, y=0.90, label= "beta == 0.849", parse=T)+
  theme_bw() 
p

库（ggplot2）
df_aa34_2您是否可以编辑您的问题，以包含
dput（df_cal）
的输出，使其易于复制？此外，我想知道您是从哪里计算汇总统计数据的？是不是
Excel
？使用
Rmisc
程序包中的
SummarySE
函数可以使这变得更容易，如您的示例链接所示。@J.Con感谢
dput（）
提示。它并不漂亮，但如果你复制并粘贴到控制台中，它似乎可以工作。我仍在使用
R
计算我的汇总统计数据。我正在使用
dplyr
从7步时间序列中手动筛选出“高原”。然后，我使用一些基本函数来生成标准偏差、标准误差和95%置信区间的向量，这三种仪器在7个步骤中的每一步都是如此。然后，我对7个校准步骤中的每个步骤执行一个
row_bind（）
，然后执行一个
unique（）
，每个校准步骤只提供一个观察值。下面是我用来生成这些汇总统计数据的示例。这看起来很棒，谢谢！我喜欢您使用
rbind（）
而不是尝试使用
restrape2
或
tidyr
。我已经将
melt（）
函数与
left\u join（）
配对，以获得类似格式的数据，但您的版本涉及的代码行更少。现在包括一些额外的
annotate（）
行，以包括第二个
lm（）
输出。