Create lagged variable in unbalanced panel data in R

问题

I'd like to create a variable containing the value of a variable in the previous year within a group.

     id   date        value
1     1   1992          4.1  
2     1     NA          4.5  
3     1   1991          3.3  
4     1   1990          5.3  
5     1   1994          3.0  
6     2   1992          3.2  
7     2   1991          5.2

value_lagged should be missing when the previous year is missing within a group - either because it is the first date within a group (as in row 4, 7), or because there are year gaps in the data (as in row 5). Also, value_lagged should be missing when the current time is missing (as in row 2).

This gives:

     id   date    value    value_lagged  
1     1   1992      4.1             3.3
2     1     NA      4.5              NA
3     1   1991      3.3             5.3
4     1   1990      5.3              NA
5     1   1994      3.0              NA
6     2   1992      3.2             5.2
7     2   1991      5.2              NA

For now, in R, I use the data.table package

 DT = data.table(id    = c(1,1,1,1,1,2,2),
                 date  = c(1992,NA,1991,1990,1994,1992,1991),
                 value = c(4.1,4.5,3.3,5.3,3.0,3.2,5.2)
                )
 setkey(DT, id, date)
 DT[, value_lagged := DT[J(id, date-1), value], ]
 DT[is.na(date), value_lagged := NA, ]

It's fast but it seems somewhat error prone to me. I'd like to know if there are better alternatives using data.table, dplyr, or any other package. Thanks a lot!

In Stata, one would do:

    tsset id date
    gen value_lagged=L.value

回答1:

I'd probably tackle this using a join:

library(dplyr)

df <- data.frame(
  id = c(1, 1, 1, 1, 1, 2, 2), 
  date = c(1992, NA, 1991, 1990, 1994, 1992, 1991), 
  value = c(4.1, 4.5, 3.3, 5.3, 3.0, 3.2, 5.2)
)


last_year <- df %>% 
  filter(!is.na(date)) %>%
  mutate(date = date + 1, lagged_value = value, value = NULL)

df %>%
  left_join(last_year)
#> Joining by: c("id", "date")
#>   id date value lagged_value
#> 1  1 1992   4.1          3.3
#> 2  1   NA   4.5           NA
#> 3  1 1991   3.3          5.3
#> 4  1 1990   5.3           NA
#> 5  1 1994   3.0           NA
#> 6  2 1992   3.2          5.2
#> 7  2 1991   5.2           NA

回答2:

Using a function tlag within groups defined by id

library(dplyr)
tlag <- function(x, n = 1L, time) { 
  index <- match(time - n, time, incomparables = NA)
  x[index]
}

df %>% group_by(id) %>% mutate(value_lagged = tlag(value, 1, time = date))

回答3:

Using 1.9.5, where joins don't need keys to be set, this can be done as follows:

require(data.table) # v1.9.5+
DT[!is.na(date), value_lagged := 
         .SD[.(id = id, date = date - 1), value, on = c("id", "date")]]
#    id date value value_lagged
# 1:  1 1992   4.1          3.3
# 2:  1   NA   4.5           NA
# 3:  1 1991   3.3          5.3
# 4:  1 1990   5.3           NA
# 5:  1 1994   3.0           NA
# 6:  2 1992   3.2          5.2
# 7:  2 1991   5.2           NA

It's a variation of your idea. The trick is to use is.na() directly in i and use .SD in j instead of DT. I've used on= syntax, but the same idea can of course be done by setting keys as well. .

来源：https://stackoverflow.com/questions/25649056/create-lagged-variable-in-unbalanced-panel-data-in-r

标签

data.table

dplyr

panel-data