Support function lag in data.table #965

Activity

1. gsee commented on Nov 20, 2014

   

   gsee

   on Nov 20, 2014

   More actions

   Just as an FYI, stats::lag is a generic. And, for what it's worth, xts
   has a lag.xts() method written in C. There is also a generic quantmod::Lag
   which has 6 methods. Not that it's all that relevant, other than something
   to look at for ideas.

   On Thu, Nov 20, 2014 at 5:41 AM, Juris Breidaks [email protected]
   wrote:

   Can include in object data.table lag function?
   As example?

   lag <- function(var, k) {
   if (k > 0) {
   #bring past valeus forward k times
   return(c(rep(NA, k), head(var, -k)))
   } else {
   #bring future values backward
   return(c(tail(var, k), rep(NA, -k)))
   }
   }

   x = data.table(a=1:10, dte=sample(seq.Date(from=as.Date("2012-01-20"),
   to=as.Date("2012-01-30"), by=1),10))
   x[, L1_a:=lag(a,1)]

   —
   Reply to this email directly or view it on GitHub
   #965.
2. arunsrinivasan commented on Dec 1, 2014

   

   arunsrinivasan

   on Dec 1, 2014

   Member

   More actions

   Please read completely and follow the instructions here. Thank you.
3. arunsrinivasan

   added

   Incomplete

   on Dec 4, 2014
4. arunsrinivasan

   self-assigned this

   on Jan 6, 2015
5. arunsrinivasan

   added this to the v1.9.6 milestone on Jan 6, 2015
6. arunsrinivasan

   closed this as completedin 30d0f17on Jan 7, 2015
7. arunsrinivasan commented on Jan 7, 2015

   

   arunsrinivasan

   on Jan 7, 2015

   Member

   More actions

   Implemented shift(), which takes type = "lag" and type = "lead". The reasons are:

   1. Avoid name clashes as much as possible.
   2. These operations are related, and therefore makes sense to be arguments under a common name.
   3. We can add more types. For example, I think adding type = "lagdiff" might be useful which is equivalent to doing c(NA, diff(.)), as I see this coming up a lot on SO. For example:

   x = 1:5
   # [1] 1 2 3 4 5
   c(NA, diff(x))
   # [1] NA  1  1  1  1
   c(NA, NA, diff(x, differences=2L)
   # [1] NA NA  0  0  0
   
   # If we want both of these in a list, then we  
   # should be able to accomplish this as:
   shift(x, n=1:2, type="lagdiff")

   4. Another type is cyclic shift.