Date: Fri, 7 Jul 2006 11:44:07 -0700
Reply-To: "Huang, Ya" <Ya.Huang@AMYLIN.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: "Huang, Ya" <Ya.Huang@AMYLIN.COM>
Subject: Re: finding minimum value in most efficient manner
Content-Type: text/plain; charset="us-ascii"
Not quite clear yet. If I change the testing data to this:
data hx;
input id jobtitle $ months;
datalines;
1 Floater 3
1 Operator 6
1 Operator 1
1 Operator 4
1 Operator 12
1 Operator 9
1 Operator 2
1 Operator 6
1 Operator 3
1 Crew 8
1 Manager 30
2 Operator 5
2 Crew 2
3 Helper 1
4 Helper 10
4 Operator 56
4 Supervisor 22
4 Manager 10
;
What's the result you want?
-----Original Message-----
From: plessthanpoinohfive [mailto:plessthanpointohfive@gmail.com]
Sent: Friday, July 07, 2006 11:32 AM
To: Huang, Ya; SAS-L@LISTSERV.UGA.EDU
Subject: RE: finding minimum value in most efficient manner
This almost works.
This made up dataset doesn't show it, but several job entries will be
the same. So, a single ID might have 8 entries of OPERATOR in a row.
In our context, I'd want to sum over these for the purposes of
identifying the shortest job. If one entry of OPERATOR is only for 1
month then it would get assigned the minflag even though the next 6
entries and the one before it are also OPERATOR. The result is that all
my flags will tag OPERATOR.
Clear as mud?
But it's a great piece of code and I could create a dataset that sums
over the consecutive jobtitles and run this code, then merge back to the
original. Not efficient.
Jen
-----Original Message-----
From: Ya Huang [mailto:ya.huang@AMYLIN.COM]
Sent: Friday, July 07, 2006 1:23 PM
To: SAS-L@LISTSERV.UGA.EDU; Jennifer Sabatier
Cc: Ya Huang
Subject: Re: finding minimum value in most efficient manner
Here is one using Howard's interleave trick:
data hx;
input id jobtitle $ months;
datalines;
1 Floater 3
1 Operator 1
1 Crew 8
1 Manager 30
2 Operator 5
2 Crew 2
3 Helper 1
4 Helper 10
4 Operator 56
4 Supervisor 22
4 Manager 10
;
data hx;
set hx (in=a_) hx (in=b_);
by id;
retain mina posa;
if first.id and a_ then do; na_=0; nb_=0; mina=9999; end; if a_ then do;
na_+1; nb_=.; if months < mina then do;
mina=months;
posa=na_;
end;
end;
if b_ then do;
nb_+1;
na_=.;
if nb_= posa then minflag=1;
if nb_= posa-1 then preflag=1;
output;
end;
drop mina posa na_ nb_;
run;
proc print;
run;
id jobtitle months minflag preflag
1 Floater 3 . 1
1 Operator 1 1 .
1 Crew 8 . .
1 Manager 30 . .
2 Operator 5 . 1
2 Crew 2 1 .
3 Helper 1 1 .
4 Helper 10 1 .
4 Operator 56 . .
4 Supervis 22 . .
4 Manager 10 . .
HTH
Ya
On Fri, 7 Jul 2006 12:44:01 -0400, Jen Sabatier
<plessthanpointohfive@GMAIL.COM> wrote:
>Hello, all,
>
>I have a dataset of individual's work histories. So, there is are
>columns for id, jobtitle, months.
>
>data hx;
>input id jobtitle $ months;
>datalines;
>1 Floater 3
>1 Operator 1
>1 Crew 8
>1 Manager 30
>2 Operator 5
>2 Crew 2
>3 Helper 1
>4 Helper 10
>4 Operator 56
>4 Supervisor 22
>4 Manager 10
>.
>.
>.
>;
>
>
>So, we have unequal lengths to the actual work history. Some people
>held several jobs, some just a few, and some only one.
>
>What I need to do is locate the shortest job held for each person for
people
>who held more than one job, flag it and the jobs before and/or after
>it. I will need to examine the previous and/or prior job along with
>the shortest job. If a person held only one job then I need to flag
them seperately.
>
>I've tried to do this with an array but I can't make it work. I know
>it must be simple.
>
>Hope someone can help!
