Skip to content

coercing created variables into a new data frame using na.omit()

4 messages · Simon Pickett, Brian Ripley, Kjetil Halvorsen

Original
Simon Pickett
# 
Hi,
I am an R newbie and one thing I am having trouble with binding variables that 
I have created within one data frame into a new data frame when using 
na.omit(). To illustrate this problem I will give the example I am working on 
and the approah I have been using:-
data.frame1<-filepath....
attach(data.frame1)
#create a new variable using a function
new.variable<-rep(1,length(weight3))
for (x in 1:length(new.variable)) 
{f<-((((age1[x]-7)*(weight[x]-mw))+((age2[x]-7)*(weight2[x]-mw))+((age3[x]-7)*
(weight3[x]-mw)))/(((age1[x]-7)^2)+((age2[x]-7)^2)+((age3[x]-7)^2))); 
new.variable[x]<-f}
#then bind it into the existing old data frame
data.frame2<-cbind(data.frame1,newvariable)
rm(dat.frame1)
attach(data.frame2)
#everything o.k. so far but now the problem part... I basically want to remove 
all the rows with NA in the new data frame including corresponding rows in the 
new variable
data.frame3<-na.omit(data.frame2)
rm(data.frame2)
attach(data.frame3)
length of new.variable has not changed but the length of all the other 
variables in data.frame2 has?
Could someone please provide an explanation or an alternative route if 
possible?
Any suggestions much appreciated,
Thankyou, Simon Pickett

Simon Pickett
Centre for Ecology and Conservation Biology
University of Exeter in Cornwall
Tremough Campus
Penryn 
Cornwall
TR10 9EZ UK
Tel: 01326371852
Brian Ripley
# 
I don't know if you can read your message, but I find it exceedingly 
difficult and there seem to be several typos.  Please use the space and 
return keys ... and only send a message once.

You problem is perhaps that you are not looking at the data frame, but at 
the variable in the workspace.  attach()ing data frames is convenient but 
error-prone (as you have found).  rm(new.variable) should solve this, but 
it is better to cultivate a different style.  For example

with(data.frame1, {
# commands to create value
data.frame1$new.variable <- value
})
data.frame3 <- na.omit(data.frame1)

I think too that the creation of the value can be vectorized simply, 
generalizing something like

value <- (age1 - 7)*(weight - mw)
On Fri, 12 Aug 2005, sp219 wrote:

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Kjetil Halvorsen
    

    

      
      #
    

  


  

      
Prof Brian Ripley wrote:

        

            

      
      
      
    

        
That cannot possible work, as assignment within with is local to
with's environment. I have used superassigmnent for this (<<-), but that 
cannot possible
be a good style?

Look at the following:

 > test <- data.frame( a=1:5, b=1:5)
 > test
  a b
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5
 > with(test, test$c <- 1:5)
 > test
  a b
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5
 > with(test, test$c <<- 1:5)
 > test
  a b c
1 1 1 1
2 2 2 2
3 3 3 3
4 4 4 4
5 5 5 5

So what is the best style her?

Kjetil

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        
-- 

Kjetil Halvorsen.

Peace is the most effective weapon of mass construction.
               --  Mahdi Elmandjra

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Brian Ripley
    

    

      
      #
    

  


  

      
On Fri, 12 Aug 2005, Kjetil Brinchmann Halvorsen wrote:

        

            

      
      
      
    

        
No, it's just a sketch of a style.

            

      
      
      
    

        
I intended that the changed object be returned: so for example

test <- with(test, {test$c <- 1:5; test})

does work.  What I really meant to write (and had tested) could be 
sketched as

value <- with(data.frame1, {
# commands to create value
})
data.frame1$new.variable <- value
data.frame3 <- na.omit(data.frame1)

but cut-and-paste got two lines out of order.