cases
allows to distinguish several cases defined logical
conditions. It can be used to code these cases into a vector. The
function can be considered as a multi-condition generalization of
ifelse
.
cases(...,check.xor=c("warn","stop","ignore"),
.default=NA,.complete=FALSE,
check.na=c("warn","stop","ignore"),
na.rm=TRUE)
If it is called with logical expressions as ... arguments,
cases
returns a factor, if it is called with
assignment expressions the function returns a vector with the
same mode as the results of the "assigned" expressions
and with the same length as the logical conditions.
A sequence of logical expressions or assignment expressions containing logical expressions as "right hand side".
character (either "warn"
, "stop"
, or "ignore"
)
or logical; if TRUE
or equal to "stop"
or
"warn"
,
cases
checks whether the case
conditions are mutually exclusive. If this is not
satisfied and check.xor
equals "warn"
(the default), a warning is shown,
otherwise an error exception is raised.
a value to be used for unsatisfied conditions.
logical, if TRUE
an additional factor level is
created for the unsatisfied conditions.
character (either "warn"
, "stop"
, or "ignore"
)
or logical; if TRUE
or equal to "stop"
or
"warn"
, cases
checks, whether any of the case
conditions evaluates to NA
.
If that case, if check.na
is TRUE
or equals
"stop"
an error exception is raised, while if check.na
equals "warn"
(the default) a warning is shown.
a logical value; how to handle NA
s (if they do not
already lead to an error exception). If FALSE
if any of the
conditions evaluates to NA
, the corresponding value of the
result vector is NA
. If TRUE
(the default), the
resulting vector or factor is NA
only for instances where all
conditions result in NA
.
There are two distinct ways to use this function. Either the
function can be used to construct a factor that represents
several logical cases or it can be used to conditionally
evaluate an expression in a manner similar to ifelse
.
For the first use, the ...
arguments have to be a series of
logical expressions. cases
then returns a factor
with as many levels as logical expressions given as
...
arguments. The resulting factor will attain its
first level if the first condition is TRUE, otherwise it will attain its
second level if the second condition is TRUE, etc.
The levels will be named after the conditions or, if name tags are
attached to the logical expressions, after the tags of the expressions.
Not that the logical expressions all need to evaluate to logical vectors
of the same length, otherwise an error condition is raised.
If .complete
is TRUE
then an additional factor level is
created for the conditions not satisfied for any of the cases.
For the second use, the ...
arguments have to be a series
of assignment expression of the type <expression> <- <logical expression>
or <logical expression> -> <expression>
. For cases
in which the first logical expression is TRUE, the result of first expression that
appears on the other side of the assignment operator become elements of the
vector returned by cases
, for cases in which the second logical expression is TRUE,
the result of the second expression that appears on the other side
of the assignment operator become elements of the
vector returned by cases
, etc.
For cases that do not satisfy any of the given conditions the value of
the .default
argument is used.
Note that the logical expressions also here all need to evaluate to logical
vectors of the same length. The expressions on the other side of the
assignment operator should also be either vectors of the same length
and mode or should scalars of the same mode, otherwise unpredictable
results may occur.
# Examples of the first kind of usage of the function
#
df <- data.frame(x = rnorm(n=20), y = rnorm(n=20))
df <- df[do.call(order,df),]
(df <- within(df,{
x1=cases(x>0,x<=0)
y1=cases(y>0,y<=0)
z1=cases(
"Condition 1"=x<0,
"Condition 2"=y<0,# only applies if x >= 0
"Condition 3"=TRUE
)
z2=cases(x<0,(x>=0 & y <0), (x>=0 & y >=0))
}))
xtabs(~x1+y1,data=df)
dd <- with(df,
try(cases(x<0,
x>=0,
x>1,
check.xor=TRUE)# let's be fussy
)
)
dd <- with(df,
try(cases(x<0,x>=0,x>1))
)
genTable(range(x)~dd,data=df)
# An example of the second kind of usage of the function:
# A construction of a non-smooth function
#
fun <- function(x)
cases(
x==0 -> 1,
abs(x)> 1 -> abs(x),
abs(x)<=1 -> x^2
)
x <- seq(from=-2,to=2,length=101)
plot(fun(x)~x)
# Demo of the new .default and .complete arguments
x <- seq(from=-2,to=2)
cases(a = x < -1,
b = x > 1,
.complete = TRUE)
cases(x < -1,
x > 1,
.complete = TRUE)
cases(1 <- x < -1,
3 <- x > 1,
.default = 2)
threshhold <- 5
d <- c(1:10, NaN)
d1 <- cases(
d > threshhold -> 1,
d <= threshhold -> 2
)
d2 <- cases(
is.na(d) -> 0,
d > threshhold -> 1,
d <= threshhold -> 2
)
# Leads to missing values because some of the conditions result in missing
# even though they could be 'captured'
d3 <- cases(
is.na(d) -> 0,
d > threshhold -> 1,
d <= threshhold -> 2,
na.rm=FALSE
)
d4 <- cases(
is.na(d) -> 0,
d > threshhold +2 -> 1,
d <= threshhold -> 2,
na.rm=FALSE
)
cbind(d,d1,d2,d3,d4)
cases(
d > threshhold,
d <= threshhold
)
cases(
is.na(d),
d > threshhold,
d <= threshhold
)
cases(
d > threshhold,
d <= threshhold,
.complete=TRUE
)
cases(
d > threshhold + 2,
d <= threshhold,
.complete=TRUE
)
Run the code above in your browser using DataLab