Recodes a numeric vector, character vector, or factor
according to simple recode specifications. Recode
is an alias for recode
that avoids name clashes
with packages, such as Hmisc, that have a recode
function.
recode(var, recodes, as.factor, as.numeric=TRUE, levels,
to.value="=", interval=":", separator=";")Recode(...)
a recoded vector of the same length as var
.
numeric vector, character vector, or factor.
character string of recode specifications: see below.
return a factor; default is TRUE
if
var
is a factor, FALSE
otherwise.
if TRUE
(the default), and as.factor
is FALSE
,
then the result will be coerced to numeric
if all values in the result can represent numbers (contain only numerals,
minus signs, etc.).
an optional argument specifying the order of the levels in the returned factor; the default is to use the sort order of the level names.
The operator to separate old from new values, "="
by default; some other possibilities: "->"
, "~"
, "~>"
.
Cannot include the interval operator (by default :
) or the separator string (by default, ;
),
so, e.g., by default ":=>"
is not allowed. The discussion
in Details assumes the default "="
. Use a non-default to.value
if
factor levels contain =
.
the operator used to denote numeric intervals, by default ":"
. The discussion in Details assumes the default ":"
. Use a non-default interval
if factor levels contain :
.
the character string used to separate recode specifications, by default ";"
. The discussion in Details assumes the default ";"
. Use a non-default separator
if factor levels contain ;
.
arguments to be passed to recode
.
The factor levels may not contain the character strings in to.value
(by default "="
), interval (by default ":"
), or separator
(by default ";"
).
John Fox jfox@mcmaster.ca
Recode specifications appear in a character string, separated by default by
semicolons (see the examples below), each of the form input=output
(where =
may be replaced by a non-default value of the
to.value
argument, e.g., input -> output
). Spaces may be
used for clarity.
If an input value satisfies more than one specification,
then the first (from left to right) applies.
If no specification is satisfied, then the input value is carried
over to the result. NA
is allowed on input and output.
Several recode specifications are supported:
For example, 0=NA
.
For example, c(7, 8, 9) = 'high'
.
For example, 7:9 = 'C'
. The special values lo
and hi
may appear in a range. For example, lo:10=1
. Note: :
is
not the R sequence operator. In addition, you may not use :
with the c
function within a recode specification, so for example c(1, 3, 5:7)
will cause an error. The :
is the default value of the recode
interval
operator; a non-default value may be specified.
else
everything that does not fit a previous specification.
For example, else = NA
. Note that else
matches all otherwise
unspecified values on input, including NA
, and if present should appear last among the recode specifications.
Character data and factor levels on the left-hand side of a recode specification must be quoted. Thus,
e.g., c(a, b, c) = 'low'
is not allowed, and should be c('a', 'b', 'c') = 'low'
.
Similarly, the colon is reserved for numeric data, and, e.g., c('a':'c') = 'low'
is not allowed.
If the var
argument is a character variable with (some) values that are character representations of numbers, or a factor
with (some) levels that are numbers (e.g., '12'
or '-2'
), then these too must be quoted
and cannot be used with colons (e.g., '15':'19' = '15 to 19'
is not allowed, and could be
specified as c('15', '16', '17', '18', '19') = '15 to 19'
, assuming that all values are
the character representation of whole numbers).
If all of the output values are numeric, and if as.factor
is
FALSE
, then a numeric result is returned; if var
is a factor,
then by default so is the result.
Fox, J. and Weisberg, S. (2019) An R Companion to Applied Regression, Third Edition, Sage.
x <- rep(1:3, 3)
x
recode(x, "c(1, 2) = 'A';
else = 'B'")
Recode(x, "1~2 -> ':=1' // 3 -> ';=2'", to.value="->",
interval="~", separator="//")
Run the code above in your browser using DataLab