[R] convert data frame of values into correlation matrix

[David Winsemius]

On Jan 30, 2010, at 2:11 PM, Juliet Hannah wrote:

> Hi Group,
>
> Consider a data frame like this:
>
> mylabel1 <- rep(c("A","B","C"),each=3)
> mylabel2 <- rep(c("A","B","C"),3)
> corrs <- c(1,.8,.7,.8,1,.7,.7,.7,1)
> myData <- data.frame(mylabel1,mylabel2,corrs)
>
> myData
>
>  mylabel1 mylabel2 corrs
> 1        A        A   1.0
> 2        A        B   0.8
> 3        A        C   0.7
> 4        B        A   0.8
> 5        B        B   1.0
> 6        B        C   0.7
> 7        C        A   0.7
> 8        C        B   0.7
> 9        C        C   1.0
>
> I would like to find a general way to get this matrix from the above  
> dataframe.
>
> corrmat <- matrix(corrs,nrow=3,byrow=TRUE)
> row.names(corrmat) <- c("A","B","C")
> colnames(corrmat) <- c("A","B","C")
> corrmat
>    A   B   C
> A 1.0 0.8 0.7
> B 0.8 1.0 0.7
> C 0.7 0.7 1.0
>
> The solution I have is the one above where I rearrange the data so
> that I can just use matrix() on one
> of the columns. I am looking for a solution in which I don't have to  
> do this.

Possibilities:

 >> reshape(myData, timevar="mylabel2", direction="wide",  
v.names="corrs", idvar="mylabel1")[,2:4]
   corrs.A corrs.B corrs.C
1     1.0     0.8     0.7
4     0.8     1.0     0.7
7     0.7     0.7     1.0
#---------
 > unstack(myData, corrs ~ mylabel2 )
     A   B   C
1 1.0 0.8 0.7
2 0.8 1.0 0.7
3 0.7 0.7 1.0
#-------
library(reshape)
mdat <- melt(myData)
cast(mdat, mylabel1 ~ mylabel2 )

   mylabel1   A   B   C
1        A 1.0 0.8 0.7
2        B 0.8 1.0 0.7
3        C 0.7 0.7 1.0


-- .

David Winsemius, MD
Heritage Laboratories
West Hartford, CT