Date: Wed, 6 Jun 2001 17:20:43 -0700
Reply-To: "Friberg, Ingrid (DHS-DCDC)" <IFriberg@DHS.CA.GOV>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: "Friberg, Ingrid (DHS-DCDC)" <IFriberg@DHS.CA.GOV>
Subject: Re: Follow-up to my summary question!
Content-Type: text/plain; charset="iso-8859-1"
The goal is to have the longest possible strings. And I guess I really don't
care whether or not it is what comes after or before that determines the
order, I just thought (from looking by hand) that it was conditioned on the
previous one, but I guess that might not be true.
-----Original Message-----
From: Ian Whitlock [mailto:WHITLOI1@WESTAT.COM]
Sent: Wednesday, June 06, 2001 4:36 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: Re: Follow-up to my summary question!
Ingrid,
Unfortunately, I missed the fact that date on ID=0 was probably wrong. I
treated it as 9/9/81 so the ** setup meets the condidition that no dates are
interchanged. Here is the ** setup that I intended.
**
2 9/9/81 B
2x 9/9/81 B
0 9/9/81 A
1 9/9/81 A
3 9/11/81 A
4 9/11/81 B
The point here is that there is no good metric for what is better. Is it
better to have one longer chain or two medium chains. If you make the change
in the renegade date the rest of my examples should be clearer. I also trie
to point out that it was not only the past records that might determine the
best arrangement, but sometimes records not yet read could influence the
decision. Please reconsider my examples in light of the date change.
Thanks to Nat Woodoing for catching the original error and pointing out that
I copied it.
Ian Whitlock
-----Original Message-----
From: Friberg, Ingrid (DHS-DCDC) [mailto:IFriberg@dhs.ca.gov]
Sent: Wednesday, June 06, 2001 5:53 PM
To: Ian Whitlock; 'SAS-L@LISTSERV.UGA.EDU'
Subject: RE: Follow-up to my summary question!
Hi Ian,
As of the data I have collected so far, there are only two on a single date,
but that isn't to say that it could never happen. The first rule is that
you can never change the order of the dates. The dates are fixed, so the **
version could NEVER happen.
In that situation, I want to get the order to be:
0 9/8/91 A
1 9/9/81 A
2 9/9/81 B
2x 9/9/81 B
4 9/11/81 B
3 9/11/81 A
because this way there are three Bs in a row.
0 9/8/91 A
1 9/9/81 A
2 9/9/81 B
2x 9/9/81 B
3 9/11/81 A
4 9/11/81 B
5 9/13/81 B
If this were the situation, then I would want the result to be:
0 9/8/91 A
1 9/9/81 A
2 9/9/81 B
2x 9/9/81 B
4 9/11/81 B
3 9/11/81 A
5 9/13/81 B
because I want to condition the series based on the ones which came
previously instead of those which come afterwards. I'm trying to avoid a
situation where you end up with a pair of singletons just because the
computer automatically sticks the A before the B without looking to see what
comes beforehand.
Does this make any more sense??
Ingrid
-----Original Message-----
From: Ian Whitlock [mailto:WHITLOI1@WESTAT.com]
Sent: Wednesday, June 06, 2001 10:47 AM
To: Friberg, Ingrid (DHS-DCDC); SAS-L@LISTSERV.UGA.EDU
Subject: RE: Follow-up to my summary question!
Ingrid,
I am not sure that I understand the rules yet, but it is an interesting
game. Suppose you had
0 9/8/91 A
1 9/9/81 A
2 9/9/81 B
2x 9/9/81 B
3 9/11/81 A
4 9/11/81 B
Now there are two arrangements
0 9/8/91 A
1 9/9/81 A
2 9/9/81 B
2x 9/9/81 B
4 9/11/81 B
3 9/11/81 A
and
**
2 9/9/81 B
2x 9/9/81 B
0 9/8/91 A
1 9/9/81 A
3 9/11/81 A
4 9/11/81 B
Is one to be preferred over the other? What if I added that the next
record is
5 9/13/81 B
Would you change your answer if the next two are
6 9/13/81 A
7 9/13/81 A
Your examples show only two toxins A and B. Can there be many? Your
examples show no more than 3 records with the same date. Can there be many
more?
Perhaps more motivation will reveal an answer.
Ian Whitlock <whitloi1@westat.com>
-----Original Message-----
From: Friberg, Ingrid (DHS-DCDC) [mailto:IFriberg@DHS.CA.GOV]
Sent: Tuesday, June 05, 2001 7:04 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: Re: Follow-up to my summary question!
I like the idea of just doing a double sort but that always puts A before B
and I want it to choose which goes first based on the previous one.
0 9/8/91 A
1 9/9/81 A
2 9/9/81 B
3 9/11/81 A
4 9/11/81 B
I want it to realize that 4 has to come before 3 because 2 is B!
Does this make sense??
Ingrid
Ingrid Friberg, MHS
Epidemiologist
Infant Botulism Treatment and Prevention Program
California Department of Health Services
2151 Berkeley Way, Room 506
Berkeley, CA 94704
Phone: 510-540-2646; Fax: 510-540-3205
email: ifriberg@dhs.ca.gov
