r

Relative frequencies of multiple variables in R

apologise for the very basic question but I am trying to calculate relative frequencies (%) of variables within a column grouped by another column. Example data:

df <- data.frame(Case = c("A","B","B","B","A","B","A","A","B","B","B"),
                          Status = c("Y","Y","Y","N","Y","ND","ND","N","Y","N","ND"))

I know how to calculate the absolute frequencies, so that the data will look like this

  Case N ND Y
1    A 1  1 2
2    B 2  2 3

This I can achieve, for instance with

newdf <- dcast(df, Case ~ Status,
               value.var = "Status", fun.aggregate = length)

But how can I calculate the percentages of cases A and B for each of the columns separately? Desired output would be something like:

  Case N ND Y
1    A 33.3  33.3 40
2    B 66.6  66.6 60

Note that each column has a different number of observations. Subsequently, I will use the output to plot frequencies of cases A and B for each status separately with facet.

All the answers I was able to find were dealing with counting absolute values only. So all help is very much appreciated!

Solution

  • 1) Base R table gets the frequencies and proportions gets those.

    round(100 * proportions(table(df), 2) , 1)
##     Status
## Case    N   ND    Y
##    A 33.3 33.3 40.0
##    B 66.7 66.7 60.0

    2) crosstable Try crosstable for a different layout.

    library(crosstable)

crosstable(df, by = "Case")
## # A tibble: 3 × 5
##   .id    label  variable A          B         
##   <chr>  <chr>  <chr>    <chr>      <chr>     
## 1 Status Status N        1 (33.33%) 2 (66.67%)
## 2 Status Status ND       1 (33.33%) 2 (66.67%)
## 3 Status Status Y        2 (40.00%) 3 (60.00%)

    3) gmmodels This package has the CrossTable funtion. The output is somewhat large so I have omitted it.

    library(gmodels)

with(df, CrossTable(Case, Status, prop.r = FALSE, prop.t = FALSE, prop.chisq = FALSE))

    4) descr This package features a CrossTable function based on the one in gmodels. It also has a plot method which produces a mosaic plot from an object of class "CrossTable" .

    library(descr)

tab <- with(df, CrossTable(Case, Status, prop.r = FALSE, prop.t = FALSE, prop.chisq = FALSE))
tab
##    Cell Contents 
## |-------------------------|
## |                       N | 
## |           N / Col Total | 
## |-------------------------|
##
## ======================================
##          Status
## Case         N      ND       Y   Total
## --------------------------------------
## A            1       1       2       4
##          0.333   0.333   0.400        
## --------------------------------------
## B            2       2       3       7
##          0.667   0.667   0.600        
## --------------------------------------
## Total        3       3       5      11
##          0.273   0.273   0.455        
## ======================================

plot(tab, color = 2:3)

    screenshot