How to create a 2x2 bar chart - next to the paired bar in R

Question

I have the following data:

  Method    Metric        E0        E1        E2        E4
1     M1 Precision 0.9661017 0.9622642 1.0000000 0.9655172
2     M2 Precision 0.5330000 0.5350000 0.3780000 0.2140000
3     M1    Recall 0.9736842 0.9736842 0.9473684 0.9473684
4     M2    Recall 1.0000000 1.0000000 1.0000000 0.6670000

What I want to do is create the above data in the following figure:

In this sketch, the bar height does not reflect the above data.

How to do it? Presumably without the use of external libraries such as ggplot or lattice.

Answer 1

Here you go:

Import

dat <- read.table(text="Method  Metric  E0  E1  E2  E4
M1  Precision   0.9661017   0.9622642   1   0.9655172
M2  Precision   0.533   0.535   0.378   0.214
M1  Recall  0.9736842   0.9736842   0.9473684   0.9473684
M2  Recall  1   1   1   0.667",header=TRUE)

Define a layout and several colors:

layout(matrix(c(1,2,5,3,4,5),nrow=2,byrow = TRUE))
#     [,1] [,2] [,3]
#[1,]    1    2    5
#[2,]    3    4    5
barcols <- c("red","blue")

Generate some graphs:

sapply(3:6, 
  function(x) {
    bp <- barplot(matrix(dat[,x],nrow=2,byrow=TRUE),beside=TRUE,col=barcols)
    title(main=names(dat[x]))
    axis(1,at=colMeans(bp),c("M1","M2"),lwd=0,lwd.tick=1)
    abline(h=0)
  }
)

Add legend:

plot(NA,xlim=c(0,1),ylim=c(0,1),ann=FALSE,axes=FALSE)
legend(0,0.6,c("Precision","Recall"),fill=barcols,cex=1.5)

Result:

Answer 2

I understand that you are trying to avoid external libraries, but ggplotwas designed to make this easy.

library(ggplot2)
library(reshape2)

gg <- melt(df,id=1:2)
ggplot(gg) +
  geom_bar(aes(x=Method, y=value, fill=Metric), stat="identity",
           position="dodge")+facet_wrap(~variable)