Don, that will still count duplicate values of b not in a, the count will
just be off by one. You can just ignore those afterwards though:
count1=. (#@[ {. <:@#/.~@,)
Or you can filter before counting:
count2=. ([:<:[:#/.~,#~[e.~,)
ts=: 6!:2,7!:2@] NB. time & space
a=: >: i.100000
b=: ?10000000$200000
ts'a count1 b'
0.7250168435948 300930496
ts'a count2 b'
0.1365689536928 117441472
In this case pre-filtering seems superior.
On Wed, Dec 11, 2013 at 4:34 PM, Don Guinn <[email protected]> wrote:
> Here is another way.
>
> <:#/.~a,b
>
> This produces a zero count for values of b not in a.
>
>
> On Wed, Dec 11, 2013 at 11:30 AM, Ni Bo <[email protected]> wrote:
>
> > Thank you. Will try it later on the data.
> >
> > Nick
> > Am 11.12.2013 20:16 schrieb "Kenneth Lettow" <[email protected]>:
> >
> > > How about
> > >
> > > a=: 1 2 3 4 5 6
> > > b=: 14 3 2 5 6 2 1 1 2 3 4 2 3
> > > +/"1 (a=/b)
> > > 2 4 3 1 1 1
> > > {:"1 /:~(~.,#)/.~(#a)-.~a i. b
> > > 2 4 3 1 1 1
> > > NB. make it a verb
> > > doit=: 4 : '{:"1 /:~(~.,#)/.~(#x)-.~x i. y'
> > > a doit b
> > > 2 4 3 1 1 1
> > > NB. larger args
> > > a=: >: i.100000
> > > b=: ?100000000$200000
> > > a doit b
> > > 516 504 497 519 505 501 498 495 485 495 503 534 500 508 482 496 495 512
> > 518
> > > 558 484 491 487 527 497 510 490 477 491 474 506 475 457 461 426 ...
> > > ts=: 6!:2,7!:2@] NB. time & space
> > > ts'a doit b'
> > > 8.0017 3.84198e9
> > >
> > >
> > >
> > >
> > > On Wed, Dec 11, 2013 at 12:11 PM, Ni Bo <[email protected]>
> wrote:
> > >
> > > > Hi,
> > > >
> > > > I'm using the data off the kaggle competition to learn some j.
> > > >
> > > > This simple code works for these short strings:
> > > > a=:1 2 3 4 5 6
> > > > b=:14 3 2 5 6 2 1 1 2 3 4 2 3
> > > > +/"1 (a=/b)
> > > > 2 4 3 1 1 1
> > > >
> > > > But if I try it for longer strings #a 250 and #b 1 000 000 I get out
> of
> > > > memory.
> > > >
> > > > Can you please help me with a better way?
> > > >
> > > > Nick
> > > >
> ----------------------------------------------------------------------
> > > > For information about J forums see
> http://www.jsoftware.com/forums.htm
> > > >
> > >
> > >
> > >
> > > --
> > > *Kenneth Lettow*
> > > *Director of Web Analytics and Market Intelligence*
> > > *ThomasNet.com*
> > > Five Penn Plaza, 8th Floor
> > > New York, NY 10001
> > > *: [email protected]
> > > *(*: 212.290.8713
> > > www.thomasnet.com
> > > ----------------------------------------------------------------------
> > > For information about J forums see http://www.jsoftware.com/forums.htm
> > >
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
> >
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
