[R] Barchart, Pareto
John Kane
jrkrideau at yahoo.ca
Mon Dec 10 17:17:11 CET 2007
--- "Kapoor, Bharat " <k at chicagogsb.edu> wrote:
> Hello
>
> Well I am relatively new so some of these issues may
> not fall under the subject that I have used.
>
> 1. How do I do a Pareto.
You might want to have a look at the qcc package. See
http://www.stat.unipg.it/~luca/Rnews_2004-1-pag11-17.pdf
for some information
Following is the approach
> I took.
>
> My data looks like this
> df2_9
> Reaason.for.failure Frequency
>
> 1 Phy Conn 1
>
> 2 Power failure 3
>
> 3 Server software 29
>
> 4 Server hardware 2
>
> 5 Server out of mem 32
>
> 6 Inadequate bandwidth 1
>
>
> I modified the data using:
> > df <-
> df2_9[order(df2_9$Frequency,decreasing=TRUE),]
> > x$Percent <- 100*(x$Frequency)/(sum(x$Frequency))
>
> ReasonCode Frequency ROF Percent
>
> 5 OOM 32 Out of Momeory 47.058824
>
> 3 SS 29 Server Software 42.647059
>
> 2 PF 3 Power failure 4.411765
>
> 4 SH 2 Server h/w 2.941176
>
> 1 PC 1 Physical Conn 1.470588
>
> 6 IB 1 Insuff b/w 1.470588
>
>
>
> barplot(x$Percent, space=0, names.arg=x$ReasonCode,
> main="Pareto", legend.text=x$ReasonCode, ylab="%
> error")
>
>
>
> Here are my questions:
>
> 1. I could not print the data in ROF column below
> the bar charts, How to get full labels for each bar
> insted I createda new coumn ReasonCode just to fit
> below the bars.
Given the lenght of the names you may have a problem.
?mtext is the obvious way but the names are too long.
Have a look at
http://finzi.psych.upenn.edu/R/Rhelp02a/archive/80007.html
for rotating the labels or use the staxlab function in
plotrix (see below)
> 2. Can I color code each bar differently
>
> 3. How to set limit on the y-axis - by deafult
> it shows 40, I want it show Y-axis till 50?
library(plotrix)
bp <- barplot(x$Percent, space=0,
names.arg=x$ReasonCode, main="Pareto",
legend.text=x$ReasonCode, ylab="% error", col=1:5,
ylim=c(0,50), xaxt="n")
staxlab(side=1, 1:6, x[,3])
> 4. How to display value of each bar inside or on
> top of it?
?text
More information about the R-help
mailing list