[R] reshape and split

[Gabor Grothendieck]

Patrick Hausmann <c18g <at> zfn.uni-bremen.de> writes:

: 
: Dear R-users,
: 
: I am trying to reshape the DF "dat2" in the "long" format,
: but can't figure out how to use the "split"-option:
: 
: > dat2
:    a.1995.z  b.1995.z  a.1996.z                    var
: 1 100.00000 100.00000 100.00000 Neue Anlagen insgesamt
: 2  40.09904  23.60890  40.88960      Neue Ausrüstungen
: 3  59.90096  76.39110  59.11040            Neue Bauten
: 
: This should be the result:
: 
: region	time	wz	value	var
: a	1995	z	100	Neue Anlagen insgesamt
: a	1995	z	40.09904	Neue Ausrüstungen
: a	1995	z	59.90096	Neue Bauten
: b	1995	z	100	Neue Anlagen insgesamt
: b	1995	z	23.60890	Neue Ausrüstungen
: b	1995	z	76.39110	Neue Bauten
: a	1996	z	100	Neue Anlagen insgesamt
: a	1996	z	40.88960	Neue Ausrüstungen
: a	1996	z	59.11040	Neue Bauten
: 
: # Not working, sure "[.]" is missing
: reshape(dat2, dir="long", varying=1:3, split=list(regexp="[0-9][a-z]",
: include=T))


Its easiest to just code times= and v.names= arguments directly rather
than use split= to try to guess them (at which point reworking it into
the form you want is straight forward):
 
nm3 <- names(dat2)[1:3]
dat2.long <- reshape(dat2, dir = "long", 
		varying = list(nm3), times = nm3, v.names = "value")

# rework into desired form
with(dat2.long, 
	data.frame(
		region = substr(time,1,1),
		time = substr(time,3,6), 
		wz = substr(time,8,8), 
		value = value, 
		var = var
	)
)