Split xts object by events
r, split, xts
Solution
Use `cumsum(ee)` to create your grouping variable, then just call `split`. You have to make a small change to the output of `cumsum`, since you want the `TRUE` values to be the last observation in the group (rather than the first).
split(xs, c(0,head(cumsum(ee),-1)))
split(xs,rev(cumsum(rev(ee)))) # grouping factors reversed
Problem
I have an xts object with a handful of event indicators. I need to split it by certain events, so that all entries from a given event up to the next event will be saved in the same `xts`, eventually creating a list of `xts` objects, with each containing the event as the last entry with no other events of the same type. an example: ``` ts = as.Date(Sys.Date()-99:0) e1 = numeric(100);e1[10*1:10]=1 e2 = numeric(100);e2[15*1:6]=1 y = 1:100 # just a sample content xs = as.xts(cbind(e1,e2,y),order.by=ts) ee = e1*e2==1 # the event in which both e1 and e2 are 1, should happen at 30,60,90 # here should be splitting function that gets xs and ee as parameters # and should return a list of 4 xts: the first with the entries 1 through 30, # the second with entries 31 to 60, the third with entries 61 to 90, and the last # with entries 91 to 100 ``` Your suggestions will be greatly appreciated.