Date:         Thu, 16 Jun 2011 11:29:22 -0500
Reply-To:     Joe Matise <snoopy369@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Joe Matise <snoopy369@GMAIL.COM>
Subject:      Re: PROC SUMMARY calculating _TYPE_.
Comments: To: "Data _null_," <iebupdte@gmail.com>
Comments: cc: Richard Read Allen <peakstat@wispertel.net>
In-Reply-To:  <BANLkTimj+Md_R=GPy9VeRD9QQZOFMz_OWA@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1

Ah, gotcha.  I've always loved using binary, ever since high school way back
when... math teams and quickly multiplying 01010110x01100111 :)

-Joe

On Thu, Jun 16, 2011 at 11:25 AM, Data _null_; <iebupdte@gmail.com> wrote:

> On Thu, Jun 16, 2011 at 10:19 AM, Joe Matise <snoopy369@gmail.com> wrote:
> > Not sure where you're getting at here.
>
> I'm just trying to gather information from folks with math skills.
>
> I thought I would get a bigger response.  But if I think about from
> looking at the code written by my "colleagues" in Pharma most of them
> don't CLASS statements in MEANS/SUMMARY they like to sort and use BY.
> Or they don't know that SUMMARY has the median and use UNIVARIATE.
>
> I see you are comfortable with powers of 2, me not so much.  Actually
> when I need to fiddle directly with the value of _TYPE_ I usually use
> the CHARTYPE option.  I could use '0101001' and not have to think too
> hard.
>
> With the introduction of the BINARY informat I can still keep my
> thinking to a minimum if I need to know as in this case 41.
>
> 2337  data _null_;
> 2338     bdg=input('0101001',binary.);
> 2339     put bdg=;
> 2340     run;
>
> bdg=41
>
>
>
> Thanks for the reference to the paper Richard I had not seen that one.
>
>
>
> On Thu, Jun 16, 2011 at 10:19 AM, Joe Matise <snoopy369@gmail.com> wrote:
> > Not sure where you're getting at here.  B*D*G would be _TYPE_=41
> (1+8+32).
> >  I think there's a way to determine it programmatically but I usually
> just
> > do the math by hand, or even just do the proc means and look if i'm
> feeling
> > particularly lazy and it's not too many lines :)
> >
> > -Joe
> >
> > On Thu, Jun 16, 2011 at 10:13 AM, Data _null_; <iebupdte@gmail.com>
> wrote:
> >>
> >> Say for
> >>
> >> CLASS A B C D E F G;
> >>
> >> How would YOU determine the value of _TYPE_ for the B*D*G conbinations.
> >>
> >> I know TYPES B*D*G; will output the records of interest but I want to
> >> know how YOU would calculate _TYPE_.
> >
> >
>