检查组中的任何日期是否在 r 中该组的特定时间间隔内

Question

我想创建一个新变量来指示 visit_date 是否在为 id

列出的任何日期范围内

我已经使用这段代码进行逐行比较，但我想扩展它并将一个 id 的所有行与为该 id 列出的间隔的所有行进行比较

df <- df %>%
  group_by(id) %>%
  mutate(between_any = ifelse((visit_date >= start & visit_date <= end), 1))

我也曾尝试创建一个区间变量并在变异之前使用 crossing(visit_date, interval)，但是我无法让 crossing 为日期对象工作。

这是一些示例数据：

df <- data.frame(id = c("a","a","a","a","a","b","b","b"),
                 visit_date = c("2001-08-22","2001-09-21","2001-10-30","2001-11-10","2001-12-20","2002-12-22", "2003-04-30","2003-05-10"),
                 start = c(NA,"2001-09-21",NA,"2001-11-10",NA,"2002-12-22", "2003-04-30",NA),
                 end = c(NA, "2001-11-01",NA,"2001-11-10",NA,"2002-12-22","2003-06-01",NA))

> df
id visit_date    start        end
a 2001-08-22       <NA>       <NA>
a 2001-09-21 2001-09-21 2001-11-01
a 2001-10-30       <NA>       <NA>
a 2001-11-10 2001-11-10 2001-11-10
a 2001-12-20       <NA>       <NA>
b 2002-12-22 2002-12-22 2002-12-22
b 2003-04-30 2003-04-30 2003-06-01
b 2003-05-10       <NA>       <NA>

我想要的输出如下：

id visit_date      start       end   between_any
a 2001-08-22       <NA>       <NA>      0
a 2001-09-21 2001-09-21 2001-11-01      1
a 2001-10-30       <NA>       <NA>      1
a 2001-11-10 2001-11-10 2001-11-10      1
a 2001-12-20       <NA>       <NA>      0
b 2002-12-22 2002-12-22 2002-12-22      1
b 2003-04-30 2003-04-30 2003-06-01      1
b 2003-05-10       <NA>       <NA>      1

提前致谢！

Answer 1

我的回答不像我想要的那样"pretty"，但它可以让你到达你想去的地方。

我先把你的日期转换成日期：

library(lubridate)
library(dplyr)
library(tibble)
library(tidyr)
library(purrr)

df <- data.frame(id = c("a","a","a","a","a","b","b","b"),
                 visit_date = c("2001-08-22","2001-09-21","2001-10-30","2001-11-10","2001-12-20","2002-12-22", "2003-04-30","2003-05-10"),
                 start = c(NA,"2001-09-21",NA,"2001-11-10",NA,"2002-12-22", "2003-04-30",NA),
                 end = c(NA, "2001-11-01",NA,"2001-11-10",NA,"2002-12-22","2003-06-01",NA)) %>%
  mutate_at(-1,as.Date)

> df
  id visit_date      start        end
1  a 2001-08-22       <NA>       <NA>
2  a 2001-09-21 2001-09-21 2001-11-01
3  a 2001-10-30       <NA>       <NA>
4  a 2001-11-10 2001-11-10 2001-11-10
5  a 2001-12-20       <NA>       <NA>
6  b 2002-12-22 2002-12-22 2002-12-22
7  b 2003-04-30 2003-04-30 2003-06-01
8  b 2003-05-10       <NA>       <NA>

接下来我为每个组创建一个间隔列表：

df_intervals <- df %>% 
  mutate_at(-1,as.Date) %>%
  filter(!is.na(start),
         !is.na(end)) %>%
  mutate(interval = start %--% end) %>%
  select(id,interval) %>%
  group_by(id)

> df_intervals
# A tibble: 4 x 2
# Groups:   id [2]
  id    interval                      
  <fct> <S4: Interval>                
1 a     2001-09-21 UTC--2001-11-01 UTC
2 a     2001-11-10 UTC--2001-11-10 UTC
3 b     2002-12-22 UTC--2002-12-22 UTC
4 b     2003-04-30 UTC--2003-06-01 UTC

最后，我根据id将区间数据加入到原始数据中，并在区间内搜索visit_date。

df_output <- df %>% as.tbl() %>%
  inner_join(df_intervals) %>%
  mutate(between_any = map2_lgl(visit_date,interval,~ .x >= int_start(.y) & .x <= int_end(.y))) %>%
  group_by(id,visit_date,start,end) %>%
  summarise(between_any = as.numeric(any(between_any)))

> df_output
# A tibble: 8 x 5
# Groups:   id, visit_date, start [8]
  id    visit_date start      end        between_any
  <fct> <date>     <date>     <date>           <dbl>
1 a     2001-08-22 NA         NA                   0
2 a     2001-09-21 2001-09-21 2001-11-01           1
3 a     2001-10-30 NA         NA                   1
4 a     2001-11-10 2001-11-10 2001-11-10           1
5 a     2001-12-20 NA         NA                   0
6 b     2002-12-22 2002-12-22 2002-12-22           1
7 b     2003-04-30 2003-04-30 2003-06-01           1
8 b     2003-05-10 NA         NA                   1

Answer 2

另一种可能是：

df %>% 
 rowid_to_column() %>%
 full_join(df %>%
            filter(!is.na(start) & !is.na(end)) %>%
            mutate(interval = interval(ymd(start), ymd(end))) %>%
            select(id, interval), by = c("id" = "id")) %>%
 group_by(rowid, id) %>%
 summarise(between_any = max(ymd(visit_date) %within% interval * 1)) %>%
 left_join(df %>%
            rowid_to_column(), by = c("rowid" = "rowid",
                                      "id" = "id")) %>%
 ungroup() %>%
 select(-rowid)
  id    between_any visit_date start      end       
  <fct>       <dbl> <fct>      <fct>      <fct>     
1 a               0 2001-11-08 <NA>       <NA>      
2 a               1 2001-09-21 2001-09-21 2001-11-01
3 a               1 2001-10-30 <NA>       <NA>      
4 a               1 2001-11-10 2001-11-10 2001-11-10
5 a               0 2001-12-20 <NA>       <NA>      
6 b               1 2002-12-22 2002-12-22 2002-12-22
7 b               1 2003-04-30 2003-04-30 2003-06-01
8 b               1 2003-05-10 <NA>       <NA>

这里它首先创建区间变量，然后根据"id"进行全连接。其次，它检查 "visit_date" 是否在每个 "id" 和 "rowid" 的任何间隔内。最后，它将结果与原始数据结合起来。

Answer 3

in_range data.table 包中的函数正是这样做的...

library(data.table)

df <- df %>%
  group_by(id) %>%
  mutate(between_any = as.numeric((inrange(visit_date, start, end))))

#> df
#  id visit_date      start        end between_any
#1  a 2001-08-22       <NA>       <NA>           0
#2  a 2001-09-21 2001-09-21 2001-11-01           1
#3  a 2001-10-30       <NA>       <NA>           1
#4  a 2001-11-10 2001-11-10 2001-11-10           1
#5  a 2001-12-20       <NA>       <NA>           0
#6  b 2002-12-22 2002-12-22 2002-12-22           1
#7  b 2003-04-30 2003-04-30 2003-06-01           1
#8  b 2003-05-10       <NA>       <NA>           1

以data.table形式...

dt <- setDT(df)      
dt[, between_any := inrange(visit_date, start, end), 
     by = id]

检查组中的任何日期是否在 r 中该组的特定时间间隔内

Check if any dates in a group are within specific time intervals for that group in r

r

lubridate

dplyr