CODEWITHSUNDEEP
rcountnumberscell
ISalvi
ISalvi

Reputation: 123

count the number of times that a number appears

I have this reduced data frame

ind;year;n
67;2016;1
76;2016;1
95;2016;2
171;2016;3
60;2017;1
73;2017;1
95;2017;3
171;2017;1
175;2017;1
60;2018;4
95;2018;7
96;2018;1
99;2018;1
171;2018;1
171;2019;2
172;2019;1
178;2019;1

and I would like to count the number of individuals that appear per year, excluding those that have appeared in previous years. In that case it would look like this:

year       n
 2016      4
 2017      3
 2018      2
 2019      2

I used this but it does not exclude those that appear in previous years

df %>%
  group_by(ind, year) %>%
  dplyr::summarise(totalcount =n())%>%
  group_by(year)%>%
  tally()

Upvotes: 3

Views: 49

Answers (2)

Ronak Shah
Ronak Shah

Reputation: 389135

We can get all ind in a list for each year. In total we create accumulating ind values in each year and to count unique ind we count of length with setdiff. 

library(dplyr)
library(purrr)

df %>%
   group_by(year) %>%
   summarise(ind = list(ind)) %>%
   mutate(total = accumulate(ind[-n()], ~unique(c(.x, .y)), .init = list()), 
           n = map2_int(ind, total, ~length(setdiff(.x, .y)))) %>%
   select(year, n)


# A tibble: 4 x 2
#   year     n
#  <int> <int>
#1  2016     4
#2  2017     3
#3  2018     2
#4  2019     2

data

df <- structure(list(ind = c(67L, 76L, 95L, 171L, 60L, 73L, 95L, 171L, 
175L, 60L, 95L, 96L, 99L, 171L, 171L, 172L, 178L), year = c(2016L, 
2016L, 2016L, 2016L, 2017L, 2017L, 2017L, 2017L, 2017L, 2018L, 
2018L, 2018L, 2018L, 2018L, 2019L, 2019L, 2019L), n = c(1L, 1L, 
 2L, 3L, 1L, 1L, 3L, 1L, 1L, 4L, 7L, 1L, 1L, 1L, 2L, 1L, 1L)), 
 class = "data.frame", row.names = c(NA, -17L))

Upvotes: 0

akrun
akrun

Reputation: 887501

Here is an option in base R

lst1 <- split(df$ind, df$year)
lst1[] <- lengths(Reduce(function(x, y) y[!x %in% y],
            split(df$ind, df$year), accumulate = TRUE))
setNames(stack(lst1)[2:1], c('year', 'n'))
#  year n
#1 2016 4
#2 2017 3
#3 2018 3
#4 2019 2

If this involves all previous 'year'

lst1 <- split(df$ind, df$year)
lst2 <- vector('list', length(lst1))
names(lst2) <- names(lst1)
lst2[[1]] <- length(lst1[[1]])
for(i in 2:length(lst1))  lst2[[i]] <- sum(!lst1[[i]] %in% 
               unlist(lst1[seq_len(i-1)]))
setNames(stack(lst2)[2:1], c('year', 'n'))
#  year n
#1 2016 4
#2 2017 3
#3 2018 2
#4 2019 2

Or an option with dplyr where we arrange by 'year', take the distinct rows (assuming that there won't be any duplicate 'ind' within a 'year'), and then use count

library(dplyr)
df %>% 
    arrange(year) %>%
    distinct(ind, .keep_all = TRUE) %>% 
    select(-n) %>%
    count(year)
# year n
#1 2016 4
#2 2017 3
#3 2018 2
#4 2019 2

data

df <- structure(list(ind = c(67L, 76L, 95L, 171L, 60L, 73L, 95L, 171L, 
175L, 60L, 95L, 96L, 99L, 171L, 171L, 172L, 178L), year = c(2016L, 
2016L, 2016L, 2016L, 2017L, 2017L, 2017L, 2017L, 2017L, 2018L, 
2018L, 2018L, 2018L, 2018L, 2019L, 2019L, 2019L), n = c(1L, 1L, 
2L, 3L, 1L, 1L, 3L, 1L, 1L, 4L, 7L, 1L, 1L, 1L, 2L, 1L, 1L)),
class = "data.frame", row.names = c(NA, 
-17L))

Upvotes: 3

Related Questions