CODEWITHSUNDEEP
rdplyr
Eric Green
Eric Green

Reputation: 7725

dplyr way to tabulate (summarise) several variables that share the same "levels"

What is the dplyr way to tabulate several variables that share the same "levels" to produce the following output?

df <- data.frame(v1 = c("sometimes", "sometimes", "rarely", "never", "often",
                        "often"),
                 v2 = c("often", "sometimes", "rarely", "never", "rarely",
                        "often"))

tab <- data.frame(cbind(table(df$v1), table(df$v2)))
names(tab) <- names(df) 
tab

#          v1 v2
#never      1  1
#often      2  2
#rarely     1  2
#sometimes  2  1

Upvotes: 2

Views: 511

Answers (2)

Jilber Urbina
Jilber Urbina

Reputation: 61154

You can use this approach:

> df %>% 
    gather(var) %>% 
    group_by(var) %>% 
    count(value) %>% 
    spread(var, n)
# A tibble: 4 x 3
  value        v1    v2
  <chr>     <int> <int>
1 never         1     1
2 often         2     2
3 rarely        1     2
4 sometimes     2     1

As pointed out by @Frank you can go straight without group_by, you can count by group just using count(), as follows:

  df %>% 
  gather %>% 
  count(key,value) %>% 
  spread(key, n)

Upvotes: 3

jdobres
jdobres

Reputation: 11957

One approach is to convert the data to "long" format, which will make easier to simply count the occurrences of your labels, then spread them into the desired format.

df.count <- df %>% 
  gather(variable, value) %>% 
  group_by(variable, value) %>% 
  count %>% 
  spread(variable, n)

  value        v1    v2
  <chr>     <int> <int>
1 never         1     1
2 often         2     2
3 rarely        1     2
4 sometimes     2     1

Of course, dplyr isn't strictly necessary:

df2 <- sapply(df, table)

This produces a named matrix, as opposed to a data frame:

          v1 v2
never      1  1
often      2  2
rarely     1  2
sometimes  2  1

And with a little more work you can turn it into a data frame:

df2 <- sapply(df, table) %>% 
  as.data.frame %>% 
  rownames_to_column(var = 'level')

Upvotes: 1

Related Questions