I have a dataset with percentages in four categories: the top two categories are “positive” and the bottom two categories are “negative”, so I want to align the border between 2 and 3, so that the zero point on all bars. (I draw pairs of bars: by the set of bars for ext.obs = 0 and one for ext.obs = 1.) Here is the data part:
structure(list(ext.obs = c(0, 0, 0, 1, 1, 1), comp = c(1, 2, 3, 1, 2, 3), `1` = c(0.00617283950617284, 0.00609756097560976, 0.0111111111111111, 0, 0, 0), `2` = c(0.154320987654321, 0.195121951219512, 0.161111111111111, 0.211180124223602, 0.392638036809816, 0.23030303030303 ), `3` = c(0.709876543209877, 0.676829268292683, 0.666666666666667, 0.745341614906832, 0.521472392638037, 0.721212121212121), `4` = c(0.12962962962963, 0.121951219512195, 0.161111111111111, 0.0434782608695652, 0.0858895705521472, 0.0484848484848485)), .Names = c("ext.obs", "comp", "1", "2", "3", "4"), row.names = c(1L, 2L, 3L, 11L, 12L, 13L), class = "data.frame")
I would like to be able to compile a matrix with this data so that I can just make barplot(datamatrix) and make it enjoyable. But I can understand nothing more than constructing the upper two categories, and then add the lower two categories using barplot(..., add=T) .
Here's the code I wrote (I actually draw 10 pairs of bars with par(mfrow=c(1, 10) ), but for(i in 1:10) ):
bar.loc <- barplot(t(as.matrix(tab3[c(i, i+10), c(5,6)])), ylim=c(-0.5, 1.0), col=my.pal[3:4], xaxt="n", yaxt="n", ylab="", xlab=components[i] ) barplot(t(as.matrix(tab3[c(i, i+10), c(4, 3)]*(-1))), add=T, col=my.pal[2:1], yaxt="n", xaxt="n", ylab="", xlab="")
You can see part of the finished product here or the image below:
Can anyone think of a more elegant way to do this?