Create a Factor Variable Using the Quantiles of a Continuous Variable

Create a factor variable using the quantiles of a continuous variable.

quantcut(x, q = 4, na.rm = TRUE, ...)

Arguments

x

Continuous variable.

q

Either a integer number of equally spaced quantile groups to create, or a vector of quantiles used for creating groups. Defaults to q=4 which is equivalent to q=seq(0, 1, by=0.25). See quantile for details.

na.rm

Boolean indicating whether missing values should be removed when computing quantiles. Defaults to TRUE.

...

Optional arguments passed to cut.

Value

Factor variable with one level for each quantile interval.

Details

This function uses quantile to obtain the specified quantiles of x, then calls cut to create a factor variable using the intervals specified by these quantiles.

It properly handles cases where more than one quantile obtains the same value, as in the second example below. Note that in this case, there will be fewer generated factor levels than the specified number of quantile intervals.

See also

cut, quantile

Author

Gregory R. Warnes greg@warnes.net

Examples

## create example data
# testonly{
set.seed(1234)
# }
x <- rnorm(1000)

## cut into quartiles
quartiles <- quantcut(x)
table(quartiles)
#> quartiles
#>    [-3.4,-0.673] (-0.673,-0.0398]  (-0.0398,0.616]      (0.616,3.2] 
#>              250              250              250              250 

## cut into deciles
deciles.1 <- quantcut(x, 10)
table(deciles.1)
#> deciles.1
#>     [-3.4,-1.21]   (-1.21,-0.849]  (-0.849,-0.538]  (-0.538,-0.285] 
#>              100              100              100              100 
#> (-0.285,-0.0398]  (-0.0398,0.193]    (0.193,0.466]    (0.466,0.761] 
#>              100              100              100              100 
#>     (0.761,1.33]       (1.33,3.2] 
#>              100              100 
# or equivalently
deciles.2 <- quantcut(x, seq(0, 1, by = 0.1))

# testonly{
stopifnot(identical(deciles.1, deciles.2))
# }

## show handling of 'tied' quantiles.
x <- round(x) # discretize to create ties
stem(x) # display the ties
#> 
#>   The decimal point is at the |
#> 
#>   -3 | 0000000000000
#>   -2 | 
#>   -2 | 000000000000000000000000000000000000000000000000
#>   -1 | 
#>   -1 | 00000000000000000000000000000000000000000000000000000000000000000000+181
#>   -0 | 
#>   -0 | 
#>    0 | 00000000000000000000000000000000000000000000000000000000000000000000+310
#>    0 | 
#>    1 | 00000000000000000000000000000000000000000000000000000000000000000000+140
#>    1 | 
#>    2 | 000000000000000000000000000000000000000000000000000000000000000
#>    2 | 
#>    3 | 00000
#> 
deciles <- quantcut(x, 10)

table(deciles) # note that there are only 5 groups (not 10)
#> deciles
#> [-3,-1)      -1       0       1   (1,3] 
#>      61     261     390     220      68 
# due to duplicates