Prevent [.data.frame drop dimensions where there is only one column - r

Question

I have a data frame demos, with n columns (depends on external input), where n = 1,2,3 ...
I want to delete certain rows, then add new columns to this data frame. When n > 1, the following code works fine, where demos.part is always an R data.frame.
demos.part <- demos[-i, ] // remove i-th row
demos.part[,"new column name"] <- as.vector(<new data>)
However when n == 1, the demos.part in the first line becomes an vector. Then the second line does not work anymore.
Of course we can hard code to fix the special case. Is there a consistent (elegant) way to remove rows from data.frame and still return a data.frame, even if the data frame has only one column?

Answer 1

Your first line, demos.part <- demos[-i, ], would only drop from a data frame to a matrix if demis.part has exactly one column:
# One column: result is a vector
> data.frame(a=letters)[1,]
[1] a
Levels: a b c d e f g h i j k l m n o p q r s t u v w x y z
# 2 cols: result is a df with 1 row
> data.frame(a=letters, b=letters)[1,]
data.frame with 1 row and 2 columns
a b
<factor> <factor>
1 a a
To see why this is, you can inspect the arguments of [.data.frame, where the default value of the drop argument depends on the number of columns:
> args(`[.data.frame`)
function (x, i, j, drop = if (missing(i)) TRUE else length(cols) ==
1)
NULL
Regardless, any time you want to prevent dropping of dimensions, simply add drop=FALSE after any indexing arguments (including intentionally blank indexing arguments; note the empty space between the two commas for the blank column index):
> data.frame(a=letters)[1, , drop=FALSE]
data.frame with 1 row and 1 column
a
<factor>
1 a
You should always use drop=FALSE when deciding how many rows/columns to select based on external input, since there is always the possibility that it will select just one row. Alternatively, use the data_frame function from the dplyr package to create a data frame with fewer weird edge cases in its behavior:
> library(dplyr)
> data_frame(a=letters)[1,]
Source: local data frame [1 x 1]
a
(chr)
1 a

Answer 2

Responding to your command about the colnames - i don't think they disappear.
Consider following code:
remove.row <- function(df,n) { as.data.frame(df[-n,]) }
#
a <- data.frame(col1=c(1,2),col2=c("A","B"))
a
class(a)
colnames(a)
#
a <- remove.row(a,1)
a
class(a)
colnames(a)
#
a <- remove.row(a,1)
a
class(a)
colnames(a)
produces:
> a
col1 col2
1 1 A
2 2 B
> class(a)
[1] "data.frame"
> colnames(a)
[1] "col1" "col2"
> #
> a <- remove.row(a,1)
> a
col1 col2
2 2 B
> class(a)
[1] "data.frame"
> colnames(a)
[1] "col1" "col2"
> #
> a <- remove.row(a,1)
> a
[1] col1 col2
<0 rows> (or 0-length row.names)
> class(a)
[1] "data.frame"
> colnames(a)
[1] "col1" "col2"