Question

I ran a program with different threads in two configurations a and b. I breakdown its timings into btime, stime, and vtime. Please see below for the data. I need to draw stacked plot as you can see below. However, I face difficulty in representing both the number of threads and configs as x-axis labels in R. Could some one help to produce a better representation of this data in terms of stacked plots in R please. Please see the figure and the R code I am using below.

Data:

config threads  btime   stime   vtime
a   2   0.08    0.32   0.09
b   2   0.32    0.19   0.16
a   4   3.72 2841.13   0.22
b   4   18.21 2865.79   5.12
a   8   5.45 2824.46   4.77
b   8   23.27 2790.14  11.89
a   16  57.63 3302.55  94.25
b   16  62.41 4041.19  82.56
a   32  119.08 3705.62 210.14
b   32  183.01 4411.14 234.17
a   64  211.51 2823.69 270.38
b   64  364.38 4091.97 387.83

R code

> barplot(t(data1[c(3:5)]), ylab="Time(seconds)", sp=c(0.1, 0.2, 1.0, 0.2, 1.0, 0.2, 1.0, 0.2, 1.0, 0.2, 1.0, 0.2),col=c("white","gray20","gray60"))
> legend("topleft",legend=c("btime","stime","vtime"), bty="n",cex=1.5 , horiz=T, adj=0.2, fill=c("white","gray20","gray60"))

Answer 1

ggplot and lattice are very helpful here, however it is useful to think of the data in a different manner: you will want to represent it as something like

config threads time_type time_value
[...]  [...]    vtime     0.03

melt from reshape accomplishes this (see tutorial http://www.statmethods.net/management/reshape.html)

Then, you can do a plot of something like…
qplot(config, time_value, data=data, group=time_type, fill=time_type, geom="barplot", facets= .~threads)

(tutorial @ http://www.r-bloggers.com/basic-introduction-to-ggplot2/)