Skip to content

Converting a data frame with values into a matrix/

7 messages · Srinivas Iyyer, Henrique Dallazuanna, John Kane +1 more

Srinivas Iyyer
# 
Dear Group, 
I have a data frame like the following:


x <- c("Mike","A",0.01)
x1 <- c("Carl","A",0.2)
x2 <- c("Gene","C",0.3)
x3 <- c("James","A",-0.3)
x4 <- c("Dough","B",0)
xx <- rbind(x,x1,x2,x3,x4)
colnames(xx)<-c("Name","Class","NES")
xx <-as.data.frame(xx)
Name Class  NES
x   Mike     A 0.01
x1  Carl     A  0.2
x2  Gene     C  0.3
x3 James     A -0.3
x4 Dough     B    0


Now I want to create a matrix with unique xx$Name on
columns and unique xx$Class as rows.  I want to fill
my choice of values (in this case 1) if data point not
available. 


xy <-
matrix(1,length(unique(xx$Class)),length(unique(xx[,1])))
colnames(xy)<-unique(xx[,1])
rownames(xy)<-unique(xx$Class)
Mike Carl Gene James Dough
A    1    1    1     1     1
C    1    1    1     1     1
B    1    1    1     1     1





I would love to have :

  Mike Carl Gene James Dough
A    0.01    0.2    1     -0.3     1
C    1    1    1     0.3     1
B    1    1    1     1    0




If I am not wrong this is called contigency or
frequeny table. 

I tried xtabs on this.
Error in Summary.factor(4L, na.rm = FALSE) : 
  sum not meaningful for factors


I tried on other data frames, it worked. BUT the
problem is it gives me 0.0000 even a value is not
available for that row and column.  So if I have data
-0.00 it is considered 0. 

I tried. drop.unused.levels = T, I did not get what I
want. I want all row.col values not available to be 1.


Is there any other trick where I map by row and column
names instead of using advanced xtabs. 

thanks
Srini
Henrique Dallazuanna
# 
Perhaps in this case:

noquote(with(xx, tapply(NES, list(Class, Name), paste)))
On 12/03/2008, Srinivas Iyyer <srini_iyyer_bio at yahoo.com> wrote:

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Srinivas Iyyer
    

    

      
      #
    

  


  

      
Hi Henrique,
Thanks for your tip. 


how can I map xx onto xy (where xy is a matrix I
created).

            

      
      
      
    

        
Mike Carl Gene James Dough
A    1    1    1     1     1
C    1    1    1     1     1
B    1    1    1     1     1


If I can map xx onto xy, I can have '1' without NAs.

Thanks
Srini

        
--- Henrique Dallazuanna <wwwhsd at gmail.com> wrote:

        

            

      
      
      
    

        
matrix(1,length(unique(xx$Class)),length(unique(xx[,1])))

            

      
      
      
    

        

      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Henrique Dallazuanna
    

    

      
      #
    

  


  

      
Try:

xy <- with(xx, tapply(NES, list(Class, Name), paste))
xy[is.na(xy)] <- 1

        
On 12/03/2008, Srinivas Iyyer <srini_iyyer_bio at yahoo.com> wrote:

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      John Kane
    

    

      
      #
    

  


  

      
The way that you have set up the data.frame is rather
unusual. It makes NES a factor which I suspect you
don't want. Do str(xx) to see what I mean

Here is a way that I think does what you want but with
the data.frame constructed in a different manner. 
Again do str(aa) to see the difference
=====================================================
aa <- data.frame(Name=c( "Mike" ,"Carl", "Gene",
"James","Dough"),
         Class=c(  "A",  "A",  "C",  "A",  "B"),
         NES=c( 0.01, 0.2, 0.3, -0.3, 0) )  
aa
         
library(reshape)
mm  <- melt(aa, id=c("Name", "Class"))
mm1  <- cast(mm, Class~Name)

aba <- mm1[,2:6]
aba[is.na(aba)] <- 1

final <- data.frame(mm1$Class, aba)
names(final) <- names(mm1) 
final

===================================================

        
--- Srinivas Iyyer <srini_iyyer_bio at yahoo.com> wrote:

        

            

      
      
      
    

        
matrix(1,length(unique(xx$Class)),length(unique(xx[,1])))

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      John Kane
    

    

      
      #
    

  


  

      
Oops, I just realised that you asked for a matrix not
a data.frame  Substituing for the last three lines of
code

final <- as.matrix(aba)
colnames(final)  <-  aa$Name
rownames(final) <- mm1$Class

should do it.

        
--- John Kane <jrkrideau at yahoo.ca> wrote:

        

            

      
      
      
    

        
=====================================================

            

      
      
      
    

        
matrix(1,length(unique(xx$Class)),length(unique(xx[,1])))

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

        
[[elided trailing spam]]

  
  


  


      

    

    

      
      

        


  

        

      Hadley Wickham
    

    

      
      #
    

  


  

          

      
      
      
    

        
You can abbreviate all this to:

final <- cast(mm, Class ~ Name, fill = 1)

Hadley