是否可以dplyr :: lag前一年而不是前一行？

问题描述

这是我的数据的一小部分

data = data.frame(
  Year = c("1994","1995","1996","1997","1998","1999","2000","2001","2004","2006","2007","2008","2009","2010","2011","2012","2013","2014","2017","2018"),RepYear = c("NA","NA","2","3","1","NA"),Id = c("A013","A013","J633","J633")
  )

   Year RepYear   Id
1  1994      NA A013
2  1995      NA A013
3  1996      NA A013
4  1997      NA A013
5  1998      NA A013
6  1999      NA A013
7  2000      NA A013
8  2001      NA A013
9  2004      NA J633
10 2006      NA J633
11 2007      NA J633
12 2008      NA J633
13 2009      NA J633
14 2010      NA J633
15 2011      NA J633
16 2012      NA J633
17 2013      NA J633
18 2014      NA J633
19 2017       2 J633
20 2017       3 J633
21 2017       1 J633
22 2018      NA J633

这就是我想用dplyr :: lag

完成的内容

   Year RepYear   Id PrevIoUsYear
1  1994      NA A013         <NA>
2  1995      NA A013         1994
3  1996      NA A013         1995
4  1997      NA A013         1996
5  1998      NA A013         1997
6  1999      NA A013         1998
7  2000      NA A013         1999
8  2001      NA A013         2000
9  2004      NA J633         <NA>
10 2006      NA J633         2004
11 2007      NA J633         2006
12 2008      NA J633         2007
13 2009      NA J633         2008
14 2010      NA J633         2009
15 2011      NA J633         2010
16 2012      NA J633         2011
17 2013      NA J633         2012
18 2014      NA J633         2013
19 2017       2 J633         2014
20 2017       3 J633         2014
21 2017       1 J633         2014
22 2018      NA J633         2017

问题是，当我像第20行和第21行那样重复年份时，因为我希望前一年= 2014，而不是上一行2017

这是我尝试过的：

data %>% arrange(Id,Year) %>%
  group_by(Id) %>%
  mutate(PrevIoUsYear = lag(Year,1)) %>%
  mutate(PrevIoUsYear = if_else(Year == lag(Year),lag(PrevIoUsYear,1),PrevIoUsYear)) %>% # Fix issue created by reapeted year
  mutate(PrevIoUsYear = if_else(Year == lag(Year),PrevIoUsYear)) # idem

但是它非常笨拙，因为显然我需要反复两次修改函数才能修复两行...

预先感谢

我。

解决方法

一种方法是仅保留Id和Year的值，然后取lag。然后，您可以将此滞后数据框加入到原始数据框中，以保持行数不变。

library(dplyr)

data %>%
  distinct(Id,Year) %>%
  group_by(Id) %>%
  mutate(prev_year = lag(Year)) %>%
  left_join(data,by = c('Year','Id'))

#   Year   Id prev_year RepYear
#1  1994 A013      <NA>      NA
#2  1995 A013      1994      NA
#3  1996 A013      1995      NA
#4  1997 A013      1996      NA
#5  1998 A013      1997      NA
#6  1999 A013      1998      NA
#7  2000 A013      1999      NA
#8  2001 A013      2000      NA
#9  2004 J633      <NA>      NA
#10 2006 J633      2004      NA
#11 2007 J633      2006      NA
#12 2008 J633      2007      NA
#13 2009 J633      2008      NA
#14 2010 J633      2009      NA
#15 2011 J633      2010      NA
#16 2012 J633      2011      NA
#17 2013 J633      2012      NA
#18 2014 J633      2013      NA
#19 2017 J633      2014       2
#20 2017 J633      2014       3
#21 2017 J633      2014       1
#22 2018 J633      2017      NA

,

其他解决方案

   df %>% 
      group_nest(Id,Year) %>% 
      group_by(Id) %>% 
      mutate(py = lag(Year)) %>% 
      unnest(data) %>% 
      ungroup()

# A tibble: 22 x 4
   Id    Year  RepYear py   
   <chr> <chr> <chr>   <chr>
 1 A013  1994  NA      NA   
 2 A013  1995  NA      1994 
 3 A013  1996  NA      1995 
 4 A013  1997  NA      1996 
 5 A013  1998  NA      1997 
 6 A013  1999  NA      1998 
 7 A013  2000  NA      1999 
 8 A013  2001  NA      2000 
 9 J633  2004  NA      NA   
10 J633  2006  NA      2004 
# ... with 12 more rows