Combining ggplot with ddply - r

Question

I have sample data test.data as follows.
income expend id
9142.7 1576.2 1
23648.75 2595 2
9014.25 156 1
4670.4 604.4 3
6691.4 3654.4 3
14425.2 66 2
8563.45 1976.2 2
2392 6 1
7915.95 619.2 3
4424.2 504.2 2
I first use ddply to get the mean income and expend for each id
library(plyr)
group<-ddply(test.data, .id,summarize, income.mean=mean(income),expend.mean=mean(expend))
Now, I use the plot function from ggplot2 to plot income.mean and expend.mean by id.
library (ggplot2)
plot.income<-qplot(id,income.mean,data=group)
plot.expend<-qplot(id,expend.mean,data=group)
While the above code runs without any error, I am looking for the efficient way to combine qplot function in ddply or vice versa. Also, if I need to combine both these plots how do I do that?
Thanks .

Answer 1

I think what you're trying to get at is going to require switching from the 'qplot' function to the 'ggplot' function. Including the graphing functions inside your 'ddply' function is not going to be very pretty, and vice versa. You're better leaving them separate, so I'm going to just focus on combining the graphs. There are two good (in my opinion) ways to do this:
Option 1: Just do both plots as separate geometries on the same 'ggplot' object. This isn't two hard to do, and works like this:
ggplot(group) + geom_point(aes(x=id, y=income.mean), colour="red") + geom_point(aes(x=id, y=expend.mean), colour="blue")
This is a fast option and gets the job done with minimal computation. However, it requires that you specify a new geometry for each column. In your sample data, this isn't an issue, but in many cases, you want to do this with code, instead of doing it by hand.
Option 2: Reshape your data to combine both sets inside of one plot. Then, we can specify groupings by coloring by the variable
library(reshape2)
plot_Data <- melt(group, id="id")
# Output of plot_Data
# id variable value
# 1 1 income.mean 6849.650
# 2 2 income.mean 12765.400
# 3 3 income.mean 6425.917
# 4 1 expend.mean 579.400
# 5 2 expend.mean 1285.350
# 6 3 expend.mean 1626.000
ggplot(plot_Data, aes(x=id, y=value, col=variable)) + geom_point()
The disadvantage of this method is that we are doing a lot more computation, so large complicated data frames may become slow to process. However, the advantage (and this is huge) is that we don't have to know what columns existed in the data frame we are plotting. Everything is sorted, colored, and plotted without our intervention, so we can use this flexibly for just about anything.
You should be able to adjust from here to suit your needs.

Answer 2

To combine both plots I had to throw in the reshape2 package to melt the data:
library(ggplot2)
library(plyr)
library(reshape2)
test.data <- read.table(text="income expend id
9142.7 1576.2 1
23648.75 2595 2
9014.25 156 1
4670.4 604.4 3
6691.4 3654.4 3
14425.2 66 2
8563.45 1976.2 2
2392 6 1
7915.95 619.2 3
4424.2 504.2 2", header=TRUE)
qplot(data=melt(ddply(test.data, .(id), colwise(mean)), id.vars="id"), x=id, y=value, colour=variable)

Answer 3

Well, your question is not very precise because we don't know what you exactly want to do. But here is a guess :
d <- read.table(textConnection("income expend id
9142.7 1576.2 1
23648.75 2595 2
9014.25 156 1
4670.4 604.4 3
6691.4 3654.4 3
14425.2 66 2
8563.45 1976.2 2
2392 6 1
7915.95 619.2 3
4424.2 504.2 2"), header=TRUE)
library(reshape2)
d2 <- melt(d, id.var="id")
ggplot(data=d2, aes(x=id,y=value)) + stat_summary(fun.y="mean", geom="bar") + facet_grid(.~variable)
Will give :