CODEWITHSUNDEEP
rdplyrdistinct
Prasanna Nandakumar
Prasanna Nandakumar

Reputation: 4335

Count non -repeats for a column

I have a dataset 

dput(Data)
    structure(list(Date = c(20170120L, 20170120L, 20170120L, 20170120L, 
    20170120L, 20170121L, 20170121L, 20170121L, 20170121L, 20170121L, 
    20170121L, 20170122L, 20170122L, 20170122L, 20170122L, 20170122L, 
    20170123L, 20170123L, 20170123L, 20170123L), Card = c(1207330L, 
    1873230L, 1556250L, 1395950L, 1395950L, 1393220L, 1058940L, 1556250L, 
    1395950L, 1395950L, 1058940L, 1207330L, 1058940L, 1700880L, 1395950L, 
    1055360L, 1395950L, 1556250L, 1207330L, 1395950L)), .Names = c("Date", 
    "Card"), class = "data.frame", row.names = c(NA, -20L))

and i am trying to get this stat

business_date     New Card
20170120             4
20170121             2
20170122             2
20170123             0

On the First day - all unique new cards(1207330,1873230,1556250,1395950) will be new cards. On the second day - all the unique cards for that second day will be compared to first day and those that are not repeated are New cards( 1393220,1058940). On the third day - need new cards that are not there in both First and second day(1700880,1055360) and so on.

Upvotes: 0

Views: 44

Answers (1)

talat
talat

Reputation: 70336

In base R this would be done by aggregateing the non-duplicated Cards per Date:

aggregate(!duplicated(df$Card), by = list(df$Date), FUN = sum)
#   Group.1 x
#1 20170120 4
#2 20170121 2
#3 20170122 2
#4 20170123 0

Or in dplyr:

library(dplyr)
df %>% 
  mutate(count = !duplicated(Card)) %>% 
  group_by(Date) %>% 
  summarise(n = sum(count))

## A tibble: 4 × 2
#      Date     n
#     <int> <int>
#1 20170120     4
#2 20170121     2
#3 20170122     2
#4 20170123     0

Upvotes: 2

Related Questions