R比較下一行的日期

Question

我在R中有這個數據框

  raw_payment_id from_bank_account        amount posted_at 
           <int> <chr>                     <dbl> <date>    
1         620691 SK660900000000062087       20.0 2018-02-25
2         618433 SK660900000000062087       10.0 2018-02-27
3         623157 SK660900000000062087       10.0 2018-03-02
4         628236 SK300900000000506871      812.  2018-03-06
5         627899 SK300900000000506871      812.  2018-03-07
6         628966 SK660900000000062087       10.0 2018-03-09

我的目標是確定是否在3天內發布了來自同一帳戶且金額相同的付款。 如果是，則將兩個付款都標記為1。這樣就可以了。

  raw_payment_id from_bank_account        amount posted_at     test 
           <int> <chr>                     <dbl> <date>        <int> 
1         620691 SK660900000000062087       20.0 2018-02-25    0
2         618433 SK660900000000062087       10.0 2018-02-27    1
3         623157 SK660900000000062087       10.0 2018-03-02    1
4         628236 SK300900000000506871      812.  2018-03-06    1
5         627899 SK300900000000506871      812.  2018-03-07    1
6         628966 SK660900000000062087       10.0 2018-03-09    0

我找不到方法，我的滯后/超前嘗試失敗了，因為銀行帳戶可能只有一筆付款。

Answer 1

library(dplyr)


df %>% 
  group_by(from_bank_account, amount) %>% 
  mutate(var = case_when(abs(as.Date(posted_at) - as.Date(lag(posted_at))) < 4 ~ 1, 
                         abs(as.Date(posted_at) - as.Date(lead(posted_at))) < 4 ~ 1,
                         TRUE ~ 0))

  raw_payment_id from_bank_account    amount posted_at    var
           <int> <fct>                 <dbl> <fct>      <dbl>
1         620691 SK660900000000062087    20. 2018-02-25    0.
2         618433 SK660900000000062087    10. 2018-02-27    1.
3         623157 SK660900000000062087    10. 2018-03-02    1.
4         628236 SK300900000000506871   812. 2018-03-06    1.
5         627899 SK300900000000506871   812. 2018-03-07    1.
6         628966 SK660900000000062087    10. 2018-03-09    0.

Answer 2

library(dplyr)

# Within each accounts, how many transactions were the same amount
tmp <- mydat %>% 
  group_by(from_bank_account, amount) %>% 
  mutate(number_of_dupes = n()) %>% 
  filter(number_of_dupes > 1) # only keep duplicates

# remove dups > 3 days apart
tmp$dup <- 0

for(i in 1:nrow(tmp)){
  acct <- tmp$from_bank_account[i]
  n    <- tmp$number_of_dupes[i]

  if(length(tmp$dup[(abs(difftime(tmp$posted_at[i],tmp$posted_at,units = "days")) < 4)
                    & (tmp$from_bank_account == acct)]) > 1){
    tmp$dup[i] <- 1
  }
}
tmp <- tmp[tmp$dup==1,]

mydat$flag_duplicate <- ifelse(mydat$raw_payment_id %in% tmp$raw_payment_id,1,0)

  raw_payment_id from_bank_account amount posted_at flag_duplicate 1 620691 SK660900000000062087 20 2018-02-25 0 2 618433 SK660900000000062087 10 2018-02-27 1 3 623157 SK660900000000062087 10 2018-03-02 1 4 628236 SK300900000000506871 812 2018-03-06 1 5 627899 SK300900000000506871 812 2018-03-07 1 6 628966 SK660900000000062087 10 2018-03-09 0