Create a table that return counts of each value for multiple variables

This is my dataframe

# Load libraries
library(data.table)
library(expss)
library(sjlabelled) # to call function as_label()

# Create dataframe
a <- data.table("b1" = c(1, 2, 2, 2),  
                "b2" = c(1, 2, 1, 1),
                "b3" = c(1, 1, 1, 1))

# Set value label
val_lab(a) = num_lab("
            1 Yes
            2 No    
")
a = as_label(a)

and it looks like this:

> a
    b1  b2  b3
1: Yes Yes Yes
2:  No  No Yes
3:  No Yes Yes
4:  No Yes Yes

I want to create a dataset that return the total occurrence counts of each value and it should look like the following:

  Category  b1  b2  b3
1:  Yes     1   3   4
2:  No      3   1   0

This might work in a similar way as the tabout command in Stata. Also, it would be great to return percentage like this

  Category  b1   b2   b3
1:  Yes    25   75   100
2:  No     75   25   0
3:  sum    100  100  100

One possibility is to use the janitor package after doing some transformations with tidyverse packages:

library(janitor)
library(dplyr)
library(tidyr)

counts <- a %>% 
  pivot_longer(everything(), values_to = "Category") %>% 
  mutate(Category = c("Yes", "No")[Category]) %>% 
  tabyl(Category, name)

Output

 Category b1 b2 b3
       No  3  1  0
      Yes  1  3  4

As for the percentages you can use the janitor adorn functions:

counts %>% 
  adorn_percentages(denominator = "col") %>% 
  adorn_totals("row") %>% 
  adorn_pct_formatting()

Note, see the ?adorn_pct_formatting for additional formatting options.

Output

 Category     b1     b2     b3
       No  75.0%  25.0%   0.0%
      Yes  25.0%  75.0% 100.0%
    Total 100.0% 100.0% 100.0%

You can use pivots and simple aggregations with tidyverse, although I would also agree with @LMc that janitor is a great package for tabular summaries.

library(tidyverse)

a |> 
  pivot_longer(everything()) |> 
  group_by(name, value) |> 
  summarise(n = n()) |> 
  mutate(p = n / sum(n)) |> 
  pivot_wider(id_cols = value, names_from = name, values_from = n, values_fill = 0)

# A tibble: 2 × 4
  value    b1    b2    b3
  <chr> <int> <int> <int>
1 No        3     1     0
2 Yes       1     3     4

Or with values_from = p instead:

# A tibble: 2 × 4
  value    b1    b2    b3
  <chr> <dbl> <dbl> <dbl>
1 No     0.75  0.25     0
2 Yes    0.25  0.75     1