Example code to split and plot a data.frame

splitAndPlot.R

if (!require(RColorBrewer)){
  install.packages("RColorBrewer")
  library(RColorBrewer)
}

### Generate a random data set                                                                                                                                                  
data <- data.frame(names=c("Type1","Type2")[as.numeric((runif(n=100)>=0.5))+1],data=rnorm(100,100,sd=25))

### Use the aggregate function to split and get the mean of the data                                                                                                            
aggregate(data$data,list(data$names),mean)

### Use the sapply and split functions to do the same thing                                                                                                                     
s <- split(data$data,list(data$names))
sapply(s,mean)
# Or the same thing in one line                                                                                                                                                 
sapply(split(data$data,list(data$names)),mean)

### Below is a function which is just a group of commands used to stop                                                                                                          
### you having to type the same code in again for another dataset                                                                                                               

plotData <- function(data,cols){
### Draw a box plot of the data                                                                                                                                                 
plot(data$data ~ data$names,col=cols,pch=20)
### Run a t-test on the split data                                                                                                                                              
pval <- t.test(data$data ~ data$names)$p.value
### Are they significantly different ?                                                                                                                                          
areSig <- c("Not Significant","Significant")[as.numeric(pval<=0.05)+1]
### Calculate the density of the data, after splitting                                                                                                                          
dens <- lapply(split(data$data,data$names),density)
### Draw an empty figure with the correct x and y limits of the data                                                                                                            
plot(1,xlim=c(0,max(sapply(dens,function(x) max(x$x)))),ylim=c(0,max(sapply(dens,function(x) max(x$y)))))
### Draw the density plots for each data type                                                                                                                                   
lapply(1:length(dens),function(x) lines(dens[[x]],col=cols[x],lwd=3))
### Add a legens                                                                                                                                                                
legend("topleft",legend=names(dens),col=cols,lwd=4)
### Add a title with the p-value and wether it is significant or not                                                                                                            
title(paste("P-value=",format.pval(pval),areSig))
}

### Draw figures in a 2 x 2 grid                                                                                                                                                
par(mfrow=c(2,2))
### Run the plotData function on the data object                                                                                                                                
plotData(data,cols=brewer.pal(8,"Dark2"))

### Make a new version of the data object, which should be significantly different, as they have different means                                                                
data <- data.frame(names=rep(c("Type3","Type4"),each=50),data=c(rnorm(50,100,sd=20),rnorm(50,50,sd=10)))
### Plot the new version of the data                                                                                                                                            
plotData(data,cols=brewer.pal(8,"Set1"))