Extract.data.frame: Extract or Replace Parts of a Data Frame

Extract.data.frameR Documentation

Extract or Replace Parts of a Data Frame


Extract or replace subsets of data frames.


## S3 method for class 'data.frame'
x[i, j, drop = ]
## S3 replacement method for class 'data.frame'
x[i, j] <- value
## S3 method for class 'data.frame'
x[[..., exact = TRUE]]
## S3 replacement method for class 'data.frame'
x[[i, j]] <- value

## S3 replacement method for class 'data.frame'
x$name <- value



data frame.

i, j, ...

elements to extract or replace. For [ and [[, these are numeric or character or, for [ only, empty or logical. Numeric values are coerced to integer as if by as.integer. For replacement by [, a logical matrix is allowed.


A literal character string or a name (possibly backtick quoted).


logical. If TRUE the result is coerced to the lowest possible dimension. The default is to drop if only one column is left, but not to drop if only one row is left.


A suitable replacement value: it will be repeated a whole number of times if necessary and it may be coerced: see the Coercion section. If NULL, deletes the column if a single column is selected.


logical: see [, and applies to column names.


Data frames can be indexed in several modes. When [ and [[ are used with a single vector index (x[i] or x[[i]]), they index the data frame as if it were a list. In this usage a drop argument is ignored, with a warning.

There is no data.frame method for $, so x$name uses the default method which treats x as a list (with partial matching of column names if the match is unique, see Extract). The replacement method (for $) checks value for the correct number of rows, and replicates it if necessary.

When [ and [[ are used with two indices (x[i, j] and x[[i, j]]) they act like indexing a matrix: [[ can only be used to select one element. Note that for each selected column, xj say, typically (if it is not matrix-like), the resulting column will be xj[i], and hence rely on the corresponding [ method, see the examples section.

If [ returns a data frame it will have unique (and non-missing) row names, if necessary transforming the row names using make.unique. Similarly, if columns are selected column names will be transformed to be unique if necessary (e.g., if columns are selected more than once, or if more than one column of a given name is selected if the data frame has duplicate column names).

When drop = TRUE, this is applied to the subsetting of any matrices contained in the data frame as well as to the data frame itself.

The replacement methods can be used to add whole column(s) by specifying non-existent column(s), in which case the column(s) are added at the right-hand edge of the data frame and numerical indices must be contiguous to existing indices. On the other hand, rows can be added at any row after the current last row, and the columns will be in-filled with missing values. Missing values in the indices are not allowed for replacement.

For [ the replacement value can be a list: each element of the list is used to replace (part of) one column, recycling the list as necessary. If columns specified by number are created, the names (if any) of the corresponding list elements are used to name the columns. If the replacement is not selecting rows, list values can contain NULL elements which will cause the corresponding columns to be deleted. (See the Examples.)

Matrix indexing (x[i] with a logical or a 2-column integer matrix i) using [ is not recommended. For extraction, x is first coerced to a matrix. For replacement, logical matrix indices must be of the same dimension as x. Replacements are done one column at a time, with multiple type coercions possibly taking place.

Both [ and [[ extraction methods partially match row names. By default neither partially match column names, but [[ will if exact = FALSE (and with a warning if exact = NA). If you want to exact matching on row names use match, as in the examples.


For [ a data frame, list or a single column (the latter two only when dimensions have been dropped). If matrix indexing is used for extraction a vector results. If the result would be a data frame an error results if undefined columns are selected (as there is no general concept of a 'missing' column in a data frame). Otherwise if a single column is selected and this is undefined the result is NULL.

For [[ a column of the data frame or NULL (extraction with one index) or a length-one vector (extraction with two indices).

For $, a column of the data frame (or NULL).

For [<-, [[<- and $<-, a data frame.


The story over when replacement values are coerced is a complicated one, and one that has changed during R's development. This section is a guide only.

When [ and [[ are used to add or replace a whole column, no coercion takes place but value will be replicated (by calling the generic function rep) to the right length if an exact number of repeats can be used.

When [ is used with a logical matrix, each value is coerced to the type of the column into which it is to be placed.

When [ and [[ are used with two indices, the column will be coerced as necessary to accommodate the value.

Note that when the replacement value is an array (including a matrix) it is not treated as a series of columns (as data.frame and as.data.frame do) but inserted as a single column.


The default behaviour when only one row is left is equivalent to specifying drop = FALSE. To drop from a data frame to a list, drop = TRUE has to be specified explicitly.

Arguments other than drop and exact should not be named: there is a warning if they are and the behaviour differs from the description here.

See Also

subset which is often easier for extraction, data.frame, Extract.


sw <- swiss[1:5, 1:4]  # select a manageable subset

sw[1:3]      # select columns
sw[, 1:3]    # same
sw[4:5, 1:3] # select rows and columns
sw[1]        # a one-column data frame
sw[, 1, drop = FALSE]  # the same
sw[, 1]      # a (unnamed) vector
sw[[1]]      # the same
sw$Fert      # the same (possibly w/ warning, see ?Extract)

sw[1,]       # a one-row data frame
sw[1,, drop = TRUE]  # a list

sw["C", ] # partially matches
sw[match("C", row.names(sw)), ] # no exact match
try(sw[, "Ferti"]) # column names must match exactly

sw[sw$Fertility > 90,] # logical indexing, see also ?subset
sw[c(1, 1:2), ]        # duplicate row, unique row names are created

sw[sw <= 6] <- 6  # logical matrix indexing

## adding a column
sw["new1"] <- LETTERS[1:5]   # adds a character column
sw[["new2"]] <- letters[1:5] # ditto
sw[, "new3"] <- LETTERS[1:5] # ditto
sw$new4 <- 1:5
sapply(sw, class)
sw$new  # -> NULL: no unique partial match
sw$new4 <- NULL              # delete the column
sw[6:8] <- list(letters[10:14], NULL, aa = 1:5)
# update col. 6, delete 7, append

## matrices in a data frame
A <- data.frame(x = 1:3, y = I(matrix(4:9, 3, 2)),
                         z = I(matrix(letters[1:9], 3, 3)))
A[1:3, "y"] # a matrix
A[1:3, "z"] # a matrix
A[, "y"]    # a matrix
stopifnot(identical(colnames(A), c("x", "y", "z")), ncol(A) == 3L,
          identical(A[,"y"], A[1:3, "y"]),
          inherits (A[,"y"], "AsIs"))

## keeping special attributes: use a class with a
## "as.data.frame" and "[" method;
## "avector" := vector that keeps attributes.   Could provide a constructor
##  avector <- function(x) { class(x) <- c("avector", class(x)); x }
as.data.frame.avector <- as.data.frame.vector

`[.avector` <- function(x,i,...) {
  r <- NextMethod("[")
  mostattributes(r) <- attributes(x)

d <- data.frame(i = 0:7, f = gl(2,4),
                u = structure(11:18, unit = "kg", class = "avector"))
str(d[2:4, -1]) # 'u' keeps its "unit"