r - 使用 ggplot2 根据 R 中的大小缩放网格中的多个饼图

Question

我想创建一个包含 9 个饼图 (3x3) 的网格，每个图表根据其大小进行缩放。 使用ggplot2和cowplot我能够创建我正在寻找的内容，但我无法进行缩放。 我只是忽略了一个功能还是应该使用另一个包？ 我还尝试了 gridExtra 包中的 grid.arrange 和 ggplot 的 facet_grid 函数，但两者都没有产生我想要的结果。

我还发现了一个使用 facet_grid 的类似问题 ( Pie charts in ggplot2 with variable pie sizes )。 不幸的是，这对我来说不起作用，因为我没有比较两个变量的所有可能结果。

这是我的示例代码:

#sample data
x <- data.frame(c("group01", "group01", "group02", "group02", "group03", "group03",
                  "group04", "group04", "group05", "group05", "group06", "group06",
                  "group07", "group07", "group08", "group08", "group09", "group09"),
                c("w","m"),
                c(8,8,6,10,26,19,27,85,113,70,161,159,127,197,179,170,1042,1230),
                c(1,1,1,1,3,3,7,7,11,11,20,20,20,20,22,22,142,142))
colnames(x) <- c("group", "sex", "data", "scale")
#I have divided the group size by the smallest group (group01, 16 people) in order to receive the scaling-variable.
#Please note that I doubled the values here for simplicity-reasons for both men and women per group (for plot-scaling only one value is needed that I calculate 
#seperately in the original data in the plot-scaling part underneath).
#In this example I am also going to use the scaling-variable as indicator of the sequence of the plots.

library(ggplot2)
library(cowplot)

#Then I create 9 pie-charts, each one containing one group and showing the quantity of men vs. women in a very simplistic style 
#(only the name of the group showing; color of each sex is explained seperately in the according text)
p1 <- ggplot(x[c(1,2),], aes("", y = data, fill = factor(sex), x$scale[1]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[1])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))
p2 <- ggplot(x[c(3,4),], aes("", y = data, fill = factor(sex), x$scale[3]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[3])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))
p3 <- ggplot(x[c(5,6),], aes("", y = data, fill = factor(sex), x$scale[5]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[5])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))
p4 <- ggplot(x[c(7,8),], aes("", y = data, fill = factor(sex), x$scale[7]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[7])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))
p5 <- ggplot(x[c(9,10),], aes("", y = data, fill = factor(sex), x$scale[9]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[9])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))
p6 <- ggplot(x[c(11,12),], aes("", y = data, fill = factor(sex), x$scale[11]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[11])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))
p7 <- ggplot(x[c(13,14),], aes("", y = data, fill = factor(sex), x$scale[13]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[13])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))
p8 <- ggplot(x[c(15,16),], aes("", y = data, fill = factor(sex), x$scale[15]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[15])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))
p9 <- ggplot(x[c(17,18),], aes("", y = data, fill = factor(sex), x$scale[17]))+
  geom_bar(width = 4, stat="identity") + coord_polar("y", start = 0, direction = 1)+
  ggtitle(label=x$group[17])+
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(),axis.line=element_blank(),axis.ticks=element_blank(),axis.text=element_blank(),plot.background = element_blank(),
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5))

#Using cowplot, I create a grid that contains my plots
plot_grid(p1,p2,p3,p4,p5,p6,p7,p8,p9, align = "h", ncol = 3, nrow = 3)

#But now I want to scale the size of the plots according to their real group size (e.g.
#group01 with 16 people vs. group09 with more than 2000 people)


#In this context, ggplot's facet_grid function produces similar results of what I want to get, 
#but since it looks at the data as a whole instead of separating groups from each other, it does not show 
#complete pie charts per group

#So is there a possibility to scale each of the 9 charts according to their group size?

这就是 plot_grid 产生的结果: pie-charts without scaling

使用 rel_widths 参数我只能调整缩放比例，但无法维持 3x3 网格。

plot_grid(p1,p2,p3,p4,p5,p6,p7,p8,p9, 
          align="h",ncol=(nrow(x)/2),          
          rel_widths = c(x$scale[1], 
                         x$scale[3], 
                         x$scale[5], 
                         x$scale[7], 
                         x$scale[9], 
                         x$scale[11], 
                         x$scale[13], 
                         x$scale[15], 
                         x$scale[17]))

这就是调整 rel_widths 的作用:

总之，我需要的是两者的混合:网格中的缩放饼图。

Answer 1

这个怎么样？

x$scale <- as.numeric(x$scale)
x$data <- as.numeric(x$data)
x$group <- factor(x$group, levels=levels(x$group)[order(x$scale[seq(1,nrow(x),2)])])

ggplot(x, aes(x=scale/2, y = data, fill = factor(sex), width=scale))+
  geom_bar(position="fill", stat="identity") + coord_polar("y")+
  facet_wrap( ~ group, nrow=3) +
  theme_classic()+theme(legend.position = "none")+
  theme(axis.title=element_blank(), axis.line=element_blank(),
        axis.ticks=element_blank(), axis.text=element_blank(),
        plot.background = element_blank(), 
        plot.title=element_text(color="black",size=10,face="plain",hjust=0.5),
        strip.background = element_blank(), 
        strip.text.x = element_text(color = "transparent") )