Skip to content
Prev 365905 / 398502 Next
Original Raw

Odd behaviour of mean() with a numeric column in a tibble

Chris Evans 
{{SIGH}} 

You are absolutely right. 

I wonder if I am losing some cognitive capacities that are needed to be part of the evolving R community. It seems to me that if a tibble is designed to be an enhanced replacement for a dataframe then it shouldn't quite so radically change things. 

I notice that the documentation on tibble says "[ Never simplifies (drops), so always returns data.frame" 
That is much less explicit than I would have liked and actually doesn't seem to be true. In fact, as you rightly say, it generally, but not quite always, returns a tibble. In fact it can be fooled into a vector of length 1.
Error in `[[.data.frame`(tmpTibble, 1, ) : 
argument "..2" is missing, with no default
# A tibble: 26 ? 1 
ID 
<chr> 
1 a 
2 b 
3 c 
4 d 
5 e 
6 f 
7 g 
8 h 
9 i 
10 j 
# ... with 16 more rows
# A tibble: 26 ? 1 
ID 
<chr> 
1 a 
2 b 
3 c 
4 d 
5 e 
6 f 
7 g 
8 h 
9 i 
10 j 
# ... with 16 more rows
Error in `[<-.data.frame`(`*tmp*`, , value = list(ID = c("a", "a", "a", : 
replacement element 3 is a matrix/data frame of 26 rows, need 1 
In addition: Warning messages: 
1: In `[<-.data.frame`(`*tmp*`, , value = list(ID = c("a", "a", "a", : 
replacement element 1 has 26 rows to replace 1 rows 
2: In `[<-.data.frame`(`*tmp*`, , value = list(ID = c("a", "a", "a", : 
replacement element 2 has 26 rows to replace 1 rows
Error: Invalid column indexes: 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26
[1] 1
int 1
Error in col[[i, exact = exact]] : 
attempt to select more than one element in vectorIndex
[1] "b"
So [[a,b]] works if a and b are legal with the dimensions of the tibble and if a is of length 1 but returns NOT a tibble but a vector of length 1 (I think), I can see that's logical but not what it says in the documentation. 

[[a]] and [[,a]] return the same result, that seems excessively tolerant to me. 

[[a,b:c]] actually returns [[a,c]] and again as a single value, NOT a tibble. 

And row subsetting/indexing has gone. 

Why create replacement for a dataframe that has no row indexing and so radically redefines column indexing, in fact redefines the whole of indexing and subsetting? 

OK. I will go to sleep now and hope to feel less dumb(ed) when I wake. Perhaps Prof. Wickham or someone can spell out a bit less tersely, and I think incompletely, than the tibble documentation does, why all this is good. 

Thanks anyway Ista, you certainly hit the issue! 

Very best all, 

Chris

        

      
      
    

        

  
  


  


  


  
  

    

  

    
      
    
    
Thread (7 messages)

  

  

    
            
        Chris Evans
        
        Odd behaviour of mean() with a numeric column in a tibble
      
      Dec 6
    
    
      
            
        Ista Zahn
        
        Odd behaviour of mean() with a numeric column in a tibble
      
      Dec 6
    
    
      
            
        Chris Evans
        
        Odd behaviour of mean() with a numeric column in a tibble
      
      Dec 6
    
    
      
            
        Jeff Newmiller
        
        Odd behaviour of mean() with a numeric column in a tibble
      
      Dec 6
    
    
      
            
        Ista Zahn
        
        Odd behaviour of mean() with a numeric column in a tibble
      
      Dec 6
    
    
      
            
        Chris Evans
        
        Odd behaviour of mean() with a numeric column in a tibble
      
      Dec 10
    
    
      
            
        Ista Zahn
        
        Odd behaviour of mean() with a numeric column in a tibble
      
      Dec 10