Date:         Wed, 3 Sep 2003 14:33:53 +0200
Reply-To:     Leslie Fisher <lfisher@METHOD-CONSULT.DE>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Leslie Fisher <lfisher@METHOD-CONSULT.DE>
Subject:      Interleaved merging/combining two datasets
Content-Type: text/plain; charset="us-ascii"

Hi,
I have problems merging two datasets (sorted by the time variables):
Dataset A
Dataset B
LNAME FNAME  DATEFROM DATETO YESNO        LNAME FNAME DATEB   VAR1 VAR2
VA33....
Smith      Jay          11.11.03           25.11.03       0
Smith      Jay          11.11.03
values............................
Presley   Elvis       13.12.02           27.12.02        1
Presley    Elvis       31.12.02
values............................

Presley   Elvis        11.09.03
values............................

Presley   Elvis        27.12.03
values............................
        etc.
What has to be done basically is a merge by NNAME, FNAME, and
DATAFROM/TO(dataset A) and DATEB (dataset B). For every YESNO=0, the
logic is: Pick and merge by LNAME+FNAME+DATEFROM/DATEB only the oldest
dataline of B to A of that individual, and for every YESNO=1, the logic
is: Pick and merge by LNAME+NMAE+DATETO/DATEB only the latest dataline
of B to A of that individual.
The result should be:
Dataset C

LNAME FNAME  DATEFROM DATETO YESNO      DATEB   VAR1 VAR2 VA33....
Smith      Jay          11.11.03           25.11.03       0
11.11.03         values.............................
Presley   Elvis       13.12.02           27.12.02        1
27.12.03         values............................
        etc.
Every individual from dataset A has definitively only one "1" or "0" in
YESNO. In the case of YESNO=0 the values of DATEFROM (dataset A) should
always correspond to the single DATEB (dataset B) value. In the case of
YESNO=1 the max value of DATETO (dataset A) should always correspond to
the last/latest DATEB (dataset B) value.
 I've tried several SQL apporaches, but got stuck.
Please help, Thanks a lot & Good luck!
Leslie