How does one merge dataframes by row name without adding a "Row.names" column?

Question

If I have two data frames, such as:

df1 = data.frame(x=1:3,y=1:3,row.names=c('r1','r2','r3'))
df2 = data.frame(z=5:7,row.names=c('r5','r6','r7'))

(

R> df1
   x y
r1 1 1
r2 2 2
r3 3 3

R> df2
   z
r5 5
r6 6
r7 7

), I'd like to merge them by row names, keeping everything (so an outer join, or all=T). This does it:

merged.df <- merge(df1,df2,all=T,by='row.names')
R> merged.df
  Row.names  x  y  z
1        r1  1  1 NA
2        r2  2  2 NA
3        r3  3  3 NA
4        r5 NA NA  5
5        r6 NA NA  6
6        r7 NA NA  7

but I want the input row names to be the row names in the output dataframe (merged.df).

I can do:

rownames(merged.df) <- merged.df[[1]]
merged.df <- merged.df[-1]

which works, but seems inelegant and hard to remember. Anyone know of a cleaner way?

Answer 1

Not sure if it's any easier to remember, but you can do it all in one step using transform.

transform(merge(df1,df2,by=0,all=TRUE), row.names=Row.names, Row.names=NULL)
#    x  y  z
#r1  1  1 NA
#r2  2  2 NA
#r3  3  3 NA
#r5 NA NA  5
#r6 NA NA  6
#r7 NA NA  7

Answer 2

From the help of merge:

If the matching involved row names, an extra character column called Row.names is added at the left, and in all cases the result has ‘automatic’ row names.

So it is clear that you can't avoid the Row.names column at least using merge. But maybe to remove this column you can subset by name and not by index. For example:

dd <- merge(df1,df2,by=0,all=TRUE) ## by=0 easier to write than row.names , 
                                   ## TRUE is cleaner than T

Then I use row.names to subset like this :

res <- subset(dd,select=-c(Row.names))
rownames(res) <- dd[,'Row.names']
  x  y  z
1  1  1 NA
2  2  2 NA
3  3  3 NA
4 NA NA  5
5 NA NA  6
6 NA NA  7