R devel - Jul 2007 - Suggestion to extend aggregate() to return multiple and/or named values

Hi all,

This is my first post to the developers list. As I understand it,  
aggregate() currently repeats a function across cells in a dataframe  
but is only able to handle functions with single value returns.  
Aggregate() also lacks the ability to retain the names given to the  
returned value. I've created an agg() function (pasted below) that is  
apparently backwards compatible (i.e. returns identical results as  
aggregate() if the function returns a single unnamed value), but is  
able to handle named and/or multiple return values. The code may be a  
little inefficient (there must be an easier way to set up the 'temp'  
data frame than to call aggregate and remove the final column), but  
I'm suggesting that something similar to this may be profitably used  
to replace aggregate entirely.

#modified aggregate command, allowing for multiple/named output values
agg=function(z,Ind,FUN,...){
	FUN.out=by(z,Ind,FUN,...)
	num.cells=length(FUN.out)
	num.dv=length(FUN.out[[1]])
	
	temp=aggregate(z,Ind,length) #dummy data frame
	temp=temp[,c(1:(length(temp)-1))] #remove last column from dummy frame
		
	for(i in 1:num.dv){
		temp=cbind(temp,NA)
		n=names(FUN.out[[1]])[i]
		names(temp)[length(temp)]=ifelse(!is.null(n),n,ifelse(i==1,'x',paste 
('x',i,sep='')))
		for(j in 1:num.cells){
			temp[j,length(temp)]=FUN.out[[j]][i]
		}
	}
	return(temp)
}

#create some factored data
z=rnorm(100) # the DV
A=rep(1:2,each=25,2) #one factor
B=rep(1:2,each=50) #another factor
Ind=list(A=A,B=B) #the factor list

aggregate(z,Ind,mean) #show the means of each cell
agg(z,Ind,mean) #should be identical to aggregate

aggregate(z,Ind,summary) #returns an error
agg(z,Ind,summary) #returns named columns

#Make a function that returns multiple unnamed values
summary2=function(x){
	s=summary(x)
	names(s)=NULL
	return(s)
}
agg(z,Ind,summary2) #returns multiple columns, default names


--
Mike Lawrence
Graduate Student, Department of Psychology, Dalhousie University

Website: http://memetic.ca

Public calendar: http://icalx.com/public/informavore/Public

"The road to wisdom? Well, it's plain and simple to express:
Err and err and err again, but less and less and less."
	- Piet Hein

Note that summaryBy in the doBy package can also do that.

library(doBy)
DF <- data.frame(z, A = Ind$A, B = Ind$B)
summaryBy(z ~ A + B, DF, FUN = summary)
summaryBy(z ~ A + B, DF, FUN = summary2)

On 7/13/07, Mike Lawrence <Mike.Lawrence at dal.ca>
wrote:
> Hi all,
>
> This is my first post to the developers list. As I understand it,
> aggregate() currently repeats a function across cells in a dataframe
> but is only able to handle functions with single value returns.
> Aggregate() also lacks the ability to retain the names given to the
> returned value. I've created an agg() function (pasted below) that is
> apparently backwards compatible (i.e. returns identical results as
> aggregate() if the function returns a single unnamed value), but is
> able to handle named and/or multiple return values. The code may be a
> little inefficient (there must be an easier way to set up the
'temp'
> data frame than to call aggregate and remove the final column), but
> I'm suggesting that something similar to this may be profitably used
> to replace aggregate entirely.
>
> #modified aggregate command, allowing for multiple/named output values
> agg=function(z,Ind,FUN,...){
>        FUN.out=by(z,Ind,FUN,...)
>        num.cells=length(FUN.out)
>        num.dv=length(FUN.out[[1]])
>
>        temp=aggregate(z,Ind,length) #dummy data frame
>        temp=temp[,c(1:(length(temp)-1))] #remove last column from dummy
frame
>
>        for(i in 1:num.dv){
>                temp=cbind(temp,NA)
>                n=names(FUN.out[[1]])[i]
>               
names(temp)[length(temp)]=ifelse(!is.null(n),n,ifelse(i==1,'x',paste
> ('x',i,sep='')))
>                for(j in 1:num.cells){
>                        temp[j,length(temp)]=FUN.out[[j]][i]
>                }
>        }
>        return(temp)
> }
>
> #create some factored data
> z=rnorm(100) # the DV
> A=rep(1:2,each=25,2) #one factor
> B=rep(1:2,each=50) #another factor
> Ind=list(A=A,B=B) #the factor list
>
> aggregate(z,Ind,mean) #show the means of each cell
> agg(z,Ind,mean) #should be identical to aggregate
>
> aggregate(z,Ind,summary) #returns an error
> agg(z,Ind,summary) #returns named columns
>
> #Make a function that returns multiple unnamed values
> summary2=function(x){
>        s=summary(x)
>        names(s)=NULL
>        return(s)
> }
> agg(z,Ind,summary2) #returns multiple columns, default names
>
>
> --
> Mike Lawrence
> Graduate Student, Department of Psychology, Dalhousie University
>
> Website: http://memetic.ca
>
> Public calendar: http://icalx.com/public/informavore/Public
>
> "The road to wisdom? Well, it's plain and simple to express:
> Err and err and err again, but less and less and less."
>        - Piet Hein
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>

bugfix already :P prior version fails when there is only one factor  
in Ind. This version also might be faster as I avoid using aggregate  
to create the dummy frame.

agg=function(z,Ind,FUN,...){
	FUN.out=by(z,Ind,FUN,...)
	num.cells=length(FUN.out)
	num.values=length(FUN.out[[1]])
	
	for(i in 1:length(Ind)){
		Ind[[i]]=unique(Ind[[i]])
	}
	temp=expand.grid(Ind)

	for(i in 1:num.values){
		temp$new=NA
		n=names(FUN.out[[1]])[i]
		names(temp)[length(temp)]=ifelse(!is.null(n),n,ifelse(i==1,'x',paste 
('x',i,sep='')))
		for(j in 1:num.cells){
			temp[j,length(temp)]=FUN.out[[j]][i]
		}
	}
	return(temp)
}


On 13-Jul-07, at 1:29 PM, Mike Lawrence wrote:
> Hi all,
>
> This is my first post to the developers list. As I understand it,  
> aggregate() currently repeats a function across cells in a  
> dataframe but is only able to handle functions with single value  
> returns. Aggregate() also lacks the ability to retain the names  
> given to the returned value. I've created an agg() function (pasted  
> below) that is apparently backwards compatible (i.e. returns  
> identical results as aggregate() if the function returns a single  
> unnamed value), but is able to handle named and/or multiple return  
> values. The code may be a little inefficient (there must be an  
> easier way to set up the 'temp' data frame than to call aggregate  
> and remove the final column), but I'm suggesting that something  
> similar to this may be profitably used to replace aggregate entirely.
>
> #modified aggregate command, allowing for multiple/named output values
> agg=function(z,Ind,FUN,...){
> 	FUN.out=by(z,Ind,FUN,...)
> 	num.cells=length(FUN.out)
> 	num.dv=length(FUN.out[[1]])
> 	
> 	temp=aggregate(z,Ind,length) #dummy data frame
> 	temp=temp[,c(1:(length(temp)-1))] #remove last column from dummy  
> frame
> 		
> 	for(i in 1:num.dv){
> 		temp=cbind(temp,NA)
> 		n=names(FUN.out[[1]])[i]
> 		names(temp)[length(temp)]=ifelse(!is.null(n),n,ifelse 
> (i==1,'x',paste('x',i,sep='')))
> 		for(j in 1:num.cells){
> 			temp[j,length(temp)]=FUN.out[[j]][i]
> 		}
> 	}
> 	return(temp)
> }
>
> #create some factored data
> z=rnorm(100) # the DV
> A=rep(1:2,each=25,2) #one factor
> B=rep(1:2,each=50) #another factor
> Ind=list(A=A,B=B) #the factor list
>
> aggregate(z,Ind,mean) #show the means of each cell
> agg(z,Ind,mean) #should be identical to aggregate
>
> aggregate(z,Ind,summary) #returns an error
> agg(z,Ind,summary) #returns named columns
>
> #Make a function that returns multiple unnamed values
> summary2=function(x){
> 	s=summary(x)
> 	names(s)=NULL
> 	return(s)
> }
> agg(z,Ind,summary2) #returns multiple columns, default names
>
>
> --
> Mike Lawrence
> Graduate Student, Department of Psychology, Dalhousie University
>
> Website: http://memetic.ca
>
> Public calendar: http://icalx.com/public/informavore/Public
>
> "The road to wisdom? Well, it's plain and simple to express:
> Err and err and err again, but less and less and less."
> 	- Piet Hein
>
>
--
Mike Lawrence
Graduate Student, Department of Psychology, Dalhousie University

Website: http://memetic.ca

Public calendar: http://icalx.com/public/informavore/Public

"The road to wisdom? Well, it's plain and simple to express:
Err and err and err again, but less and less and less."
	- Piet Hein