I find SQL's groupby or Unix's uniq is quite a useful concept.
gb=. 1 : '[: (u&.>@{. , <)&> (1,~2 ~:/\ u&.>) <;.2 ]'
u is the key verb, which returns keys for groupby.
] gb 'aaaabbaacc'
+-+----+
|a|aaaa|
+-+----+
|b|bb |
+-+----+
|a|aa |
+-+----+
|c|cc |
+-+----+
d=.'May 17';'May 18';'June 5';'July 10';'July 21'
(>@{.;#@>@{:)"1 >@{.@;: gb d
+----+-+
|May |2|
+----+-+
|June|1|
+----+-+
|July|2|
+----+-+
Maybe groupby as a conjunction would be simpler:
gbc=. 2 : '[: (u&>@{. ; v)&> (1,~2 ~:/\ u&.>) <;.2 ]'
(>@{.@;: gbc #) d
+----+-+
|May |2|
+----+-+
|June|1|
+----+-+
|July|2|
+----+-+
I'd appreciate any suggestion for improvement -- better and more
elegant approaches.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
