CODEWITHSUNDEEP
rdplyrsummarize
Omry Atia
Omry Atia

Reputation: 2443

Number of categories not equal to a specific one

I have a data frame with many categorical columns. I would like to count the number of distinct categories not equal to "bla". So for example:

> d1
# A tibble: 5 x 2
    x      y    
  <chr>  <chr>
1 yellow A    
2 green  A    
3 green  bla  
4 blue   B    
5 bla    bla

How can I modify dplyr's 

d1 %>% summarise_all(n_distinct)

to exclude the category "bla"? In this case, the answer should be 3 for column x and 2 for column y.

Upvotes: 2

Views: 69

Answers (3)

Ronak Shah
Ronak Shah

Reputation: 388982

We can use filter_all to filter the rows from all the columns and then use n_distinct to get length of unique values.

library(dplyr)

d1 %>% 
   filter_all(all_vars(. != "bla")) %>% 
    summarise_all(n_distinct)

#  x y
#1 3 2

Upvotes: 3

zx8754
zx8754

Reputation: 56149

Using base::lengths():

lengths(lapply(d1, function(i) unique(i[ i != "bla" ])))
# x y 
# 3 2

Upvotes: 1

fidelin
fidelin

Reputation: 320

Using data table

library(data.table)

d1 <- data.table(d1)

d1[!y %like% 'bla', .(count = .N, distinct = uniqueN(x)), by = .(y)]

Upvotes: 0

Related Questions