将每个矩阵列转换为子矩阵并循环遍历每个子矩阵以进行绘图

Question

我有一个数据框，其中包含以下因素：

rss<-c("rs","total","rs","total","total","total","rs","rs")
css<-c("G","S","S","Q","Q","Q","G","G")
batch<-c("5","6","6","7","7","8","7","5")
dat<-data.frame(rss,css,batch)

我想为每列绘制上述数据帧的数值矩阵，即按列将8 * 3矩阵转换为3,4 * 2子矩阵。这样我就可以遍历并绘制子矩阵数据。

newdat<-data.matrix(dat)
split.dat<-split(newdat,col(newdat))
par(mfrow=c(1,3))
for(i in split.dat){image(array(i,c(4,2)))}

有没有更好的方法来做同样的事情？也欢迎任何关于绘制image

以外的此类数据的建议

Answer 1

以下是其他一些选项，不过你的方式很好。

apply(newdat, 2, function(i) image(matrix(i, 4, 2)))

或者，使用raster代替

library(raster)
par(mfrow=c(1,3))
apply(newdat, 2, function(i) plot(raster(matrix(i, 4, 2))) # plot as raster
# or
apply(newdat, 2, function(i) plot(raster(matrix(i, 4, 2)), useRaster=FALSE))
# pictured below, uses image()

或使用persp

nr <- 4
nc <- 2
par(mfrow=c(1,3))
apply(newdat, 2, function(i)
   persp(0:nc, 0:nr, matrix(1, (nc+1), (nr+1)), zlim=c(0,2), phi=270,
       box=FALSE, col = t(matrix(i, 4, 2))))
# the color matrix here is using the integers to define colors, but you may want to adjust that.

来自levelplot的{​​{1}}：

lattice

Answer 2

我认为这解决了ggplot2的问题。但是，您的数据格式很奇怪，并且不清楚为什么要按照您描述的方式绘制数据。在我看来，有一些变量缺失，特别是与我创建的row数据框中的col和newdat相对应的变量。我建议你阅读Hadley Wickham的tidy data paper。

require(tidyr)
require(dplyr)
require(ggplot2)

rss<-c("rs","total","rs","total","total","total","rs","rs")
css<-c("G","S","S","Q","Q","Q","G","G")
batch<-c("5","6","6","7","7","8","7","5")
dat<-data.frame(rss,css,batch)

# convert into tidy data frame
dat %>% mutate(col=c(rep(1,4), rep(2,4)), row=c(1:4, 1:4)) %>%
  gather(key, value, rss:batch) -> newdat

# plot
ggplot(newdat, aes(x=col, y=row, fill=value)) + 
  geom_tile() +
  facet_wrap(~key) +
  scale_x_discrete(breaks=c(1,2), limits=c(1,2)) +
  theme_bw()

结果：