合并基于多列的两个数据集，但使时间列在分钟范围内灵活

Question

我有一个看起来像这样的数据集

id|date     |social_id     | race    | age | time        |Location
 1  04/02/19  2000001          W        29    "04:10:05"  HA
 2  04/06/20  2000002          B        22    "05:12:49"  CA
 3  04/12/20  2000021          B        26    "09:13:32"  MA
 4  08/14/20  2000026          A        29    "06:12:34"  VT

第二个数据集看起来像这样

    id2|date     |social_id     | race    | age | time| sex
      1  04/02/19  2000001          W        29    "04:30:05"  M
      2  04/06/20  2000002          B        22    "05:49:49"  F
      3  04/12/20  2000021          B        26    "10:13:32"  M
      4  08/14/20  2000026          A        29    "06:19:54"  F

请注意，除时间外，所有列都相同。我想根据这些列进行连接

日期、social_id、race_age 和时间。但是两个数据集的时间都不匹配

df3 <- df1 %>% left_join(df2,by=c("date","social_id","race","time"))

有没有一种方法可以进行多列连接，但在 45 分钟内对时间进行例外处理？时间是字符串格式，所以我通过写

来调整它

abs(difftime(as.POSIXct(strptime(df1$time,format="%H:%M:%S")), as.POSIXct(strptime(df2$time,format="%H:%M:%S")),units = "mins")) <= 45

它可以独立工作并识别时间字符串是否在 45 分钟内。当我进行合并时，如何将它们组合在一起？

structure(list(id = 1:4, date = c("4/2/2019", "4/6/2020", "4/12/2020", 
"8/14/2020"), race = c("w", "b", "b", "a"), age = c(29L, 22L, 
26L, 29L), time = structure(c(15005L, 18769L, 33212L, 22354L), class = 
"ITime")), row.names = c(NA, 
-4L), class = "data.frame")

structure(list(id2 = 1:4, date = c("4/2/2019", "4/6/2020", "4/12/2020", 
"8/14/2020"), race = c("w", "b", "b", "a"), age = c(29L, 22L, 
26L, 29L), time = structure(c(16205L, 20989L, 36812L, 22794L), class = 
"ITime")), row.names = c(NA, 
-4L), class = "data.frame")

Answer 1

我们可以使用 lubridate

中的 round_date

library(dplyr)
library(lubridate)
library(stringr)
df1 %>% 
    mutate(datetime = round_date(mdy_hms(str_c(date, time,
       sep = ' ')), '45 mins')) %>% 
    left_join(df2 %>%
       mutate(datetime = round_date(mdy_hms(str_c(date, time,
          sep = ' ')), '45 mins')), 
    by = c('datetime', 'id' = 'id2', 'race', 'age'))

合并基于多列的两个数据集，但使时间列在分钟范围内灵活

Merge two datasets based on multiple columns but make time column flexible within minute range

time

merge

r

strftime