Reordering the columns in a data frame

Problem

You want to do reorder the columns in a data frame.

Solution

  1. # A sample data frame
  2. data <- read.table(header=TRUE, text='
  3.     id weight   size
  4.      1     20  small
  5.      2     27  large
  6.      3     24 medium
  7. ')
  9. # Reorder by column number
  10. data[c(1,3,2)]
  11. #>   id   size weight
  12. #> 1  1  small     20
  13. #> 2  2  large     27
  14. #> 3  3 medium     24
  16. # To actually change `data`, you need to save it back into `data`:
  17. # data <- data[c(1,3,2)]
  20. # Reorder by column name
  21. data[c("size", "id", "weight")]
  22. #>     size id weight
  23. #> 1  small  1     20
  24. #> 2  large  2     27
  25. #> 3 medium  3     24

The above examples index into the data frame by treating it as a list (a data frame is essentially a list of vectors). You can also use matrix-style indexing, as in data[row, col], where row is left blank.

  1. data[, c(1,3,2)]
  2. #>   id   size weight
  3. #> 1  1  small     20
  4. #> 2  2  large     27
  5. #> 3  3 medium     24

The drawback to matrix indexing is that it gives different results when you specify just one column. In these cases, the returned object is a vector, not a data frame. Because the returned data type isn’t always consistent with matrix indexing, it’s generally safer to use list-style indexing, or the drop=FALSE option:

  1. # List-style indexing of one column
  2. data[2]
  3. #>   weight
  4. #> 1     20
  5. #> 2     27
  6. #> 3     24
  8. # Matrix-style indexing of one column - drops dimension to become a vector
  9. data[,2]
  10. #> [1] 20 27 24
  12. # Matrix-style indexing with drop=FALSE - preserves dimension to remain data frame
  13. data[, 2, drop=FALSE]
  14. #>   weight
  15. #> 1     20
  16. #> 2     27
  17. #> 3     24