R/introduction.md

In mabafaba/categorical: Representing and Manipulating categorical data in R

Categorical Vectors

features

Categorical Vectors are a basic vector type with the following features:

• they have a fixed set of allowed values called 'levels' (similar to levels in factors)
• they can store alternative values for each level; you can switch back and forth between alternatives.
• each record in the vector can have more than one level 'selected' to allow storing multipe response data.
• when printing categorical values, they are wrapped in single quotes (between factors that are printed with no quotes and characters that are printed with double quotes)

Example: days of the week

let's store the days of the week as integer values 1-7, but include labels in two different languages, and another alternative value that defines whether the day is on the weekend or not:

mydays <- categorical(c(1,2,1,5,6,3,7),
                                levels=1:7,
                                labels_english = c('Monday','Tuesday','Wednesday','Thursday','Friday','Saturday','Sunday'),
                                labels_german = c('Montag','Dienstag','Mittwoch','Donnerstag','Freitag','Samstag','Sonntag'),
                                day_type = c('week','week','week','week','week','weekend','weekend')
                                )

we can alternate between the original and alternative values:

mydays_en <- alternate(mydays, "labels_english")

You can see that the levels remain untouched. we can alternate back to the original levels if we don't name the alternative:

alternate(mydays_en)
#> '1' '2' '1' '5' '6' '3' '7'
#> levels: 1 2 3 4 5 6 7

let's subset the vector to only keep the weekend days:

my_weekend_days_en <- mydays_en[alternate(mydays_en, 'day_type')==categorical('weekend')]
my_weekend_days_en
#> categorical vector of length 0
#> levels: 1 2 3 4 5 6 7

categorical gives you four new basic vectors:

• categorical itself: a vector with levels (predefined categorical values in the vector, just like in factors), labels (sort of like names), and the option to define any number of alternatives values for each level. The main values can be characters, numbers or factors. Each record / item in the vector can have any number of levels selected (including none).
• three specific classes derived from it: ordinal and range

The categorical class

library(dplyr)
#> Warning: package 'dplyr' was built under R version 3.5.2
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

my_colors <- categorical(x = c("red","blue" , "red","red", "blue"), 
                         levels =  c("red", "green", "blue"),
                         hexcode = c("#FF0000","#00FF00","#0000FF"))

my_colors                
#> 'red' 'blue' 'red' 'red' 'blue'
#> levels: red green blue

you can switch to the alternative value at any time:

my_colors_as_hexcodes<-alternate(my_colors, alternative = 'hexcode')
my_colors_as_hexcodes
#> '#FF0000' '#0000FF' '#FF0000' '#FF0000' '#0000FF'
#> levels: red green blue

if you don't provide an alternative argument, the vector is reverted to the original value:

alternate(my_colors_as_hexcodes)
#> 'red' 'blue' 'red' 'red' 'blue'
#> levels: red green blue

Note that multiple 'original' values can have the same alternative; one 'original' can however not point to multiple different alternatives. So the 'original' vector should always be the most granular one.

categorical vectors with levels and alternatives

# What is your favourite colour?

my_colors <- categorical(x = c("red","blue" , "red","red", "blue"), 
                         levels =  c("red", "green", "blue"),
                         hexcode = c("#FF0000","#00FF00","#0000FF"),
                         temperature = c('warm', 'cool', 'cool'))


color_temp <- alternate(my_colors, 'temperature')

color_temp
#> 'warm' 'cool' 'warm' 'warm' 'cool'
#> levels: red green blue

alternate(color_temp)
#> 'red' 'blue' 'red' 'red' 'blue'
#> levels: red green blue

We can switch as suitable:

colour_df<-tibble(my_colors)
colour_df
#> # A tibble: 5 x 1
#>   my_colors
#>    <ctgrcl>
#> 1     'red'
#> 2    'blue'
#> 3     'red'
#> 4     'red'
#> 5    'blue'

Multiple selection

categorical vectors support multiple selection. That means that for each element of a vector, each level can be 'selected' or not selected:

# What are your favourite colour? (some people like more than one color, or even no colors at all!)



many_favourite_colours <- categorical(list(c('red'),
                                           c('blue', 'orange', 'yellow'),
                                           c('yellow', 'magenta'),
                                           c('black'),
                                           c()),
                                      levels = c('red','blue','orange','yellow','magenta','black'))

many_favourite_colours
#>  (1) 'red'  (3) 'blue' & 'orange' & 'yellow'  (2) 'yellow' & 'magenta'  (1) 'black'  (0) ''
#> levels: red blue orange yellow magenta black

Since we didn't specify the levels, they were set automatically to the unique values in the supplied vector. Note that the levels correspond to all given answers, not answer combinations:

levels(many_favourite_colours)
#> [1] "red"     "blue"    "orange"  "yellow"  "magenta" "black"

Ordinal class

# colors ordered by wavelength
my_lightwaves<-ordinal(x = c('red','infra','violet','blue','red', 'infra'),
                levels = c('violet','blue','red', 'infra'),rank = c(1,2,3,4))

my_lightwaves
#> #3-red #4-infra #1-violet #2-blue #3-red #4-infra

Ordinals as all categorical classes support multiple selection:

my_lightwaves<-ordinal(x = list(c('red','infra'),
                                'violet',
                                c('blue','red'),
                                'infra'),
                levels = c('violet','blue','red', 'infra'),rank = c(1,2,3,4))



my_lightwaves
#> (#3-red & #4-infra) #1-violet (#2-blue & #3-red) #4-infra

Intervals

mabafaba/categorical documentation built on May 3, 2020, 9 p.m.