R help - Jan 2010 - apply function with grouped columns

I have a data set of many rows and many columns in which both the rows and
the columns have associated grouping factors.

Is there a way to do what 'aggregate' does but in the other dimension?

The way I have been doing this is to use 'aggregate' on the data in the
usual way and then rotate the result and apply 'aggregate' again. This
works but is a little messy and I was wondering if there is some built-in
way to do this easier.

An example of what I have been doing is below.

#Four observations of four variables
v1 <- c(1, 8, 5, 3)
v2 <- c(5, 5, 6, 5)
v3 <- c(3, 2, 9, 4)
v4 <- c(4, 1, 1, 1)

myData <- data.frame(v1=v1, v2=v2, v3=v3, v4=v4)
myData

#The observations have either property 1 or property 2
rowFactor <- data.frame(RowTraits=factor(c(1, 2, 1, 2)))
rowFactor

#The variables have a property that is either present or absent
colFactor <- data.frame(ColTraits=factor(c(1, 1, 0, 0)))
colFactor

#Getting the means for the columns by row groups is easy
MeanByRowTraits <- aggregate (myData, rowFactor[1], mean)
MeanByRowTraits

#But now to get the means for the rows by column groups is awkward
rotateData <- data.frame(t(MeanByRowTraits[2:5]))
colnames(rotateData) <- c(levels(rowFactor[,1]))
rotateData

#This is the kind of result I want in the end
aggregate (rotateData, colFactor[1], mean)

-- 
Jack Siegrist
Graduate Program in Ecology & Evolution,
and Department of Ecology, Evolution, & Natural Resources,
Rutgers University, New Brunswick, NJ 08901

Try this:

y <- c(v1, v2, v3, v4)
rowf <- gl(2, 1, 16)
colf <- gl(2, 8, 16, labels=1:0)
dat <- data.frame(y)
aggregate(dat[1], list(Row=rowf, Col=colf), mean)

  -Peter Ehlers

Jack Siegrist wrote:
> I have a data set of many rows and many columns in which both the rows and
> the columns have associated grouping factors.
> 
> Is there a way to do what 'aggregate' does but in the other
dimension?
> 
> The way I have been doing this is to use 'aggregate' on the data in
the
> usual way and then rotate the result and apply 'aggregate' again.
This
> works but is a little messy and I was wondering if there is some built-in
> way to do this easier.
> 
> An example of what I have been doing is below.
> 
> #Four observations of four variables
> v1 <- c(1, 8, 5, 3)
> v2 <- c(5, 5, 6, 5)
> v3 <- c(3, 2, 9, 4)
> v4 <- c(4, 1, 1, 1)
> 
> myData <- data.frame(v1=v1, v2=v2, v3=v3, v4=v4)
> myData
> 
> #The observations have either property 1 or property 2
> rowFactor <- data.frame(RowTraits=factor(c(1, 2, 1, 2)))
> rowFactor
> 
> #The variables have a property that is either present or absent
> colFactor <- data.frame(ColTraits=factor(c(1, 1, 0, 0)))
> colFactor
> 
> #Getting the means for the columns by row groups is easy
> MeanByRowTraits <- aggregate (myData, rowFactor[1], mean)
> MeanByRowTraits
> 
> #But now to get the means for the rows by column groups is awkward
> rotateData <- data.frame(t(MeanByRowTraits[2:5]))
> colnames(rotateData) <- c(levels(rowFactor[,1]))
> rotateData
> 
> #This is the kind of result I want in the end
> aggregate (rotateData, colFactor[1], mean)
> 
-- 
Peter Ehlers
University of Calgary