R: losing column names when adding rows to an empty data frame

Question

I am just starting with R and encountered a strange behaviour: when inserting the first row in an empty data frame, the original column names get lost.

example:

a<-data.frame(one = numeric(0), two = numeric(0))
a
#[1] one two
#<0 rows> (or 0-length row.names)
names(a)
#[1] "one" "two"
a<-rbind(a, c(5,6))
a
#  X5 X6
#1  5  6
names(a)
#[1] "X5" "X6"

As you can see, the column names one and two were replaced by X5 and X6.

Could somebody please tell me why this happens and is there a right way to do this without losing column names?

A shotgun solution would be to save the names in an auxiliary vector and then add them back when finished working on the data frame.

Thanks

Context:

I created a function which gathers some data and adds them as a new row to a data frame received as a parameter. I create the data frame, iterate through my data sources, passing the data.frame to each function call to be filled up with its results.

Answer 1

The rbind help pages specifies that :

For ‘cbind’ (‘rbind’), vectors of zero length (including ‘NULL’) are ignored unless the result would have zero rows (columns), for S compatibility. (Zero-extent matrices do not occur in S3 and are not ignored in R.)

So, in fact, a is ignored in your rbind instruction. Not totally ignored, it seems, because as it is a data frame the rbind function is called as rbind.data.frame :

rbind.data.frame(c(5,6))
#  X5 X6
#1  5  6

Maybe one way to insert the row could be :

a[nrow(a)+1,] <- c(5,6)
a
#  one two
#1   5   6

But there may be a better way to do it depending on your code.

Answer 2

was almost surrendering to this issue.

1) create data frame with stringsAsFactor set to FALSE or you run straight into the next issue

2) don't use rbind - no idea why on earth it is messing up the column names. simply do it this way:

df[nrow(df)+1,] <- c("d","gsgsgd",4)

df <- data.frame(a = character(0), b=character(0), c=numeric(0))

df[nrow(df)+1,] <- c("d","gsgsgd",4)

#Warnmeldungen:
#1: In `[<-.factor`(`*tmp*`, iseq, value = "d") :
#  invalid factor level, NAs generated
#2: In `[<-.factor`(`*tmp*`, iseq, value = "gsgsgd") :
#  invalid factor level, NAs generated

df <- data.frame(a = character(0), b=character(0), c=numeric(0), stringsAsFactors=F)

df[nrow(df)+1,] <- c("d","gsgsgd",4)

df
#  a      b c
#1 d gsgsgd 4

Answer 3

Workaround would be:

a <- rbind(a, data.frame(one = 5, two = 6))

?rbind states that merging objects demands matching names:

It then takes the classes of the columns from the first data frame, and matches columns by name (rather than by position)

Answer 4

FWIW, an alternative design might have your functions building vectors for the two columns, instead of rbinding to a data frame:

ones <- c()
twos <- c()

Modify the vectors in your functions:

ones <- append(ones, 5)
twos <- append(twos, 6)

Repeat as needed, then create your data.frame in one go:

a <- data.frame(one=ones, two=twos)

Answer 5

One way to make this work generically and with the least amount of re-typing the column names is the following. This method doesn't require hacking the NA or 0.

rs <- data.frame(i=numeric(), square=numeric(), cube=numeric())
for (i in 1:4) {
    calc <- c(i, i^2, i^3)
    # append calc to rs
    names(calc) <- names(rs)
    rs <- rbind(rs, as.list(calc))
}

rs will have the correct names

> rs
    i square cube
1   1      1    1
2   2      4    8
3   3      9   27
4   4     16   64
> 

Another way to do this more cleanly is to use data.table:

> df <- data.frame(a=numeric(0), b=numeric(0))
> rbind(df, list(1,2)) # column names are messed up
>   X1 X2
> 1  1  2

> df <- data.table(a=numeric(0), b=numeric(0))
> rbind(df, list(1,2)) # column names are preserved
   a b
1: 1 2

Notice that a data.table is also a data.frame.

> class(df)
"data.table" "data.frame"

Answer 6

You can do this:

give one row to the initial data frame

 df=data.frame(matrix(nrow=1,ncol=length(newrow))

add your new row and take out the NAS

newdf=na.omit(rbind(newrow,df))

but watch out that your newrow does not have NAs or it will be erased too.

Cheers Agus

Answer 7

I use the following solution to add a row to an empty data frame:

d_dataset <- 
  data.frame(
    variable = character(),
    before = numeric(),
    after = numeric(),
    stringsAsFactors = FALSE)

d_dataset <- 
  rbind(
    d_dataset,
      data.frame(
        variable = "test",
        before = 9,
        after = 12,
        stringsAsFactors = FALSE))  

print(d_dataset)

variable before after  
1     test      9    12

HTH.

Kind regards

Georg

Answer 8

Instead of constructing the data.frame with numeric(0) I use as.numeric(0).

a<-data.frame(one=as.numeric(0), two=as.numeric(0))

This creates an extra initial row

a
#    one two
#1   0   0

Bind the additional rows

a<-rbind(a,c(5,6))
a
#    one two
#1   0   0
#2   5   6

Then use negative indexing to remove the first (bogus) row

a<-a[-1,]
a

#    one two
#2   5   6

Note: it messes up the index (far left). I haven't figured out how to prevent that (anyone else?), but most of the time it probably doesn't matter.