r - 根据额外因素用多边形边框覆盖 ggplot 分组图 block

Question

我有一个包含 x 和 y 位置以及两个因子列 blocknr 和 cat 的数据框:

dput(testData)
structure(list(xpos = c(2L, 8L, 5L, 8L, 1L, 4L, 5L, 1L, 8L, 4L, 
3L, 2L, 6L, 5L, 1L, 7L, 3L, 4L, 3L, 7L, 1L, 6L, 7L, 7L, 2L, 5L, 
3L, 4L, 6L, 7L, 1L, 5L, 1L, 6L, 4L, 5L, 3L, 6L, 4L, 8L, 1L, 3L, 
4L, 6L, 7L, 3L, 2L, 6L, 4L, 2L, 1L, 7L, 4L, 8L, 2L, 3L, 2L, 5L, 
8L, 2L, 8L, 3L, 3L, 5L, 6L, 7L, 1L, 5L, 6L, 4L, 2L, 6L, 7L, 1L, 
5L, 7L, 2L), ypos = c(1L, 2L, 8L, 1L, 6L, 7L, 1L, 4L, 6L, 1L, 
2L, 3L, 4L, 5L, 7L, 8L, 10L, 2L, 6L, 9L, 1L, 2L, 10L, 4L, 5L, 
6L, 3L, 5L, 9L, 3L, 9L, 10L, 3L, 7L, 8L, 2L, 5L, 6L, 3L, 4L, 
10L, 1L, 4L, 10L, 2L, 8L, 9L, 3L, 6L, 8L, 5L, 7L, 10L, 3L, 4L, 
7L, 2L, 4L, 5L, 6L, 7L, 9L, 4L, 7L, 8L, 1L, 2L, 9L, 5L, 9L, 10L, 
1L, 6L, 8L, 3L, 5L, 7L), blocknr = c(1L, 3L, 2L, 3L, 1L, 2L, 
2L, 1L, 3L, 2L, 1L, 1L, 3L, 2L, 1L, 3L, 2L, 2L, 1L, 3L, 1L, 2L, 
3L, 3L, 1L, 2L, 1L, 2L, 3L, 3L, 1L, 2L, 1L, 3L, 2L, 2L, 1L, 3L, 
2L, 3L, 1L, 1L, 2L, 3L, 3L, 2L, 1L, 3L, 2L, 1L, 1L, 3L, 2L, 3L, 
1L, 2L, 1L, 2L, 3L, 1L, 3L, 2L, 1L, 2L, 3L, 3L, 1L, 2L, 3L, 2L, 
1L, 2L, 3L, 1L, 2L, 3L, 1L), cat = structure(c(2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L), .Label = c("A", "B", "C"
), class = "factor")), .Names = c("xpos", "ypos", "blocknr", 
"cat"), row.names = c(NA, -77L), class = "data.frame")

我制作了以下 ggplot 代码来制作 2D 概览:

ggplot(data=testData, aes(x=xpos,y=ypos))+
geom_tile(aes(fill=cat), colour = "white")+
scale_fill_manual(values = c('A' = '#F8766D','C' = '#8ABF54','B' = '#C1DDA5'))+
geom_text(aes(x=xpos,y=ypos,label=blocknr),size=3)+
coord_cartesian(ylim = c(0.5, ymax + 0.5)) +
coord_cartesian(xlim = c(0.5, xmax + 0.5)) +
scale_x_continuous(breaks=seq(1,xmax,1))+
scale_y_continuous(breaks=seq(1,ymax,1))+
#geom_polygon(aes(group=blocknr))+
theme(axis.line = element_line(colour = "white"),
    panel.grid.major = element_blank(),
    panel.grid.minor = element_blank(),
    panel.border = element_blank(),
    panel.background = element_blank())

产生以下结果:

现在我想通过在它们周围绘制边框来突出显示每组 blocknrs，如下所示:

我玩过 geom_polygon、geom_path，但我找不到解决这个问题的方法。在 ggplot 中是否有一种通用的方法可以实现这一点，而无需构建算法来计算每条线的位置并将这些线添加为 geom_segment？

Answer 1

据我所知，使用标准的 ggplot2 tile 选项无法做到这一点。但是，如果将它们作为段来构建它们并没有太大的麻烦。例如

ymax <- max(testData$ypos)
xmax <- max(testData$xpos)

m <- matrix(0, nrow=ymax, ncol=xmax)
m[as.matrix(testData[,2:1])] <- testData[,3]

这里我们基本上是获取所有行/列分配数据并创建一个基本上看起来像图但我们将使用 block 号的矩阵。现在，我们将在分别遍历每一行和每一列时通过查找 block 编号的变化来扫描我们需要添加“墙”的位置。

has.breaks<-function(x) ncol(x)==2 & nrow(x)>0

hw<-do.call(rbind.data.frame, Filter(has.breaks, Map(function(i,x) 
    cbind(y=i,x=which(diff(c(0,x,0))!=0)), 1:nrow(m), split(m, 1:nrow(m)))))
vw<-do.call(rbind.data.frame, Filter(has.breaks, Map(function(i,x)
    cbind(x=i,y=which(diff(c(0,x,0))!=0)), 1:ncol(m), as.data.frame(m))))

并且您可以添加对 geom_segments 的调用以将水平和垂直墙添加到绘图中。

ggplot(data=testData, aes(x=xpos,y=ypos))+
    geom_tile(aes(fill=cat), colour = "white")+
    scale_fill_manual(values = c('A' = '#F8766D','C' = '#8ABF54','B' = '#C1DDA5'))+
    geom_text(aes(x=xpos,y=ypos,label=blocknr),size=3)+
    geom_segment(data=hw, aes(x=x-.5, xend=x-.5, y=y-.5, yend=y+.5))+
    geom_segment(data=vw, aes(x=x-.5, xend=x+.5, y=y-.5, yend=y-.5))+
    coord_cartesian(ylim = c(0.4, ymax + 0.6)) +
    coord_cartesian(xlim = c(0.4, xmax + 0.6)) +
    scale_x_continuous(breaks=seq(1,xmax,1))+
    scale_y_continuous(breaks=seq(1,ymax,1))+
    theme(axis.line = element_line(colour = "white"),
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_blank(),
        panel.background = element_blank())

给出