Question

Suppose we want to access data from a data frame by row. The examples are simplified but when ordering a data frame by row names, for example, (df[order(row.names(df)]) we use the same technique.

If the data frame has one column, we get back an atomic vector:

> df
    x1
a   x
b   y
c   z

> df[1, ] # returns atomic vector
[1] x 

If the data frame has two columns, we get back a 1-row data frame including the row name:

> df
    x1 x2
a   x  u
b   y  v
c   z  w 

> df[1, ] # returns data frame
   X1 X2
a  x  u 

I don’t understand why the same operation on the data frame yields two types of results depending on how many columns the frame has.

Answer 1

It’s because the default argument to [ is drop=TRUE.

From ?"["

drop
For matrices and arrays. If TRUE the result is coerced to the
lowest possible dimension (see the examples). This only works for
extracting elements, not for the replacement. See drop for further
details.

> dat1 <- data.frame(x=letters[1:3])
> dat2 <- data.frame(x=letters[1:3], y=LETTERS[1:3])

The default behaviour:

> dat[1, ]
     row sessionId scenarionName stepName duration
[1,]   1      1001             A    start        0

> dat[2, ]
     row sessionId scenarionName stepName duration
[1,]   2      1001             A    step1      2.2

Using drop=FALSE:

> dat1[1, , drop=FALSE]
  x
1 a

> dat2[1, , drop=FALSE]
  x y
1 a A