Ok, my J model was wrong. I just put it together mentally. Now I've created
some test data and corrected the model:
require'dates'
N =: # tkrs =: ;:'IBM CSCO DELL AAPL F X'
day1 =: 1 + day0 =: todayno 2006 01 01
'X Y' =: cocreate&.>2#a:
symbol__Y =: tkrs {~ 3 ?. # tkrs
date__X =: day1 + ?. days #~ changes =. ?. N # <: <. N %~ days
=. 365
symbol__X =: tkrs , date__X /:~ changes # tkrs
date__X =: (day0 #~ #tkrs), /:~ date__X
prod_id__X =: i.#symbol__X
map_it =: dyad define
filter =. (symbol__x e. symbol__y) *. date__x <: todayno 2006
06 15
rows =. filter # prod_id__x ,. date__x
prod_ids =. rows {.@:\:/@:|:/.~ symbol__y i. filter # symbol__x
symbol__y ,. <"0 prod_ids
)
X map_it Y
+----+--+
|IBM |83|
+----+--+
|CSCO|81|
+----+--+
|X |65|
+----+--+
I hope my need is clearer now. Can anyone help?
-Dan
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
