R:有条件地将数据从一个数据帧提取到另一个

Question

我有两个数据帧，我想有条件地从一个数据帧的一列中提取数据并将其放入另一个数据帧的新列中。
数据框 1 如下所示:

df1 <- data.frame(date.start = c("2019-06-10 11:52:00",
  "2019-06-11 11:52:00", "2019-06-12 11:51:00"), date.end =
  c("2019-06-10 11:53:00", "2019-06-11 11:53:00", "2019-06-12 11:53:00"))

数据框 2 如下所示:

df2 <- data.frame(date.start = c("2019-06-11 11:50:00",
  "2019-06-10 11:51:00", "2019-06-12 11:50:00"), date.end =
  c("2019-06-11 11:54:00", "2019-06-11 08:59:00", "2019-06-12 11:57:00"),
  day = c(1, 15, 64))

如果date.start和 date.end df.1 属于 date.start或 date.end df2 的任意行我想提取变量 day来自 df2并将其放入df1的匹配行中.
预期结果如下所示:

expected.out <- data.frame(date.start = c("2019-06-10 11:52:00", "2019-06-11 11:52:00", "2019-06-12 11:51:00"),
                           date.end = c("2019-06-10 11:53:00", "2019-06-11 11:53:00", "2019-06-12 11:53:00"),
                           day = c(15, 1, 64))

我目前有以下循环有效，但是当我在我的大数据帧(行 = 1135133)上运行它时它非常慢，我想知道是否有更快的方法来做到这一点。

for(i in 1:nrow(df1)){
  find.match <- which(df1$date.start[i] >= df2$date.start &
                        df1$date.end[i] <= df2$date.end)
  if(length(find.match) !=0){
    df1$day[i] <- df2$day[find.match]
  }
  
}

Answer 1

使用 library(fuzzyjoin)

library(tidyverse)
library(lubridate)
library(fuzzyjoin)

df1 <- data.frame(
  date.start = c("2019-06-10 11:52:00", "2019-06-11 11:52:00", "2019-06-12 11:51:00"),
  date.end = c("2019-06-10 11:53:00", "2019-06-11 11:53:00", "2019-06-12 11:53:00"), stringsAsFactors = F)

df2 <- data.frame(date.start = c("2019-06-11 11:50:00", "2019-06-10 11:51:00", "2019-06-12 11:50:00"),
                  date.end = c("2019-06-11 11:54:00", "2019-06-11 08:59:00", "2019-06-12 11:57:00"),
                  day = c(1, 15, 64), stringsAsFactors = F)

df1 <- df1 %>% 
  mutate(across(where(is.character), ymd_hms)) %>% 
  as_tibble()

df2 <- df2 %>% 
  mutate(across(where(is.character), ymd_hms)) %>% 
  as_tibble()


fuzzy_left_join(df1, df2, by = c("date.start", "date.end"), match_fun = list(`>=`, `<=`))
# A tibble: 3 x 5
  date.start.x        date.end.x          date.start.y        date.end.y            day
  <dttm>              <dttm>              <dttm>              <dttm>              <dbl>
1 2019-06-10 11:52:00 2019-06-10 11:53:00 2019-06-10 11:51:00 2019-06-11 08:59:00    15
2 2019-06-11 11:52:00 2019-06-11 11:53:00 2019-06-11 11:50:00 2019-06-11 11:54:00     1
3 2019-06-12 11:51:00 2019-06-12 11:53:00 2019-06-12 11:50:00 2019-06-12 11:57:00    64

创建于 2020-09-23 由 reprex package (v0.3.0)
不确定该方法是否快速