Sometimes, due to upstream data pipeline issues, we have to work with a version history that isn't completely up to date, but with functions that expect archives that are completely up to date, or equally as up-to-date as another archive. This function provides one way to approach such mismatches: pretend that we've "observed" additional versions, filling in these versions with NAs or extrapolated values.
Usage
epix_fill_through_version(x, fill_versions_end, how = c("na", "locf"))
Arguments
- x
An
epi_archive
- fill_versions_end
Length-1, same class&type as
x$version
: the version through which to fill in missing version history; this will be the result's$versions_end
unless it already had a later$versions_end
.- how
Optional;
"na"
or"locf"
:"na"
will fill in any missing required version history withNA
s, by inserting (if necessary) an update immediately after the current$versions_end
that revises all existing measurements to beNA
(this is only supported forversion
classes with anext_after
implementation);"locf"
will fill in missing version history with the last version of each observation carried forward (LOCF), by leaving the update$DT
alone (otherepi_archive
methods are based on LOCF). Default is"na"
.