Date:         Mon, 10 Jan 2000 22:07:26 +0100
Reply-To:     "Christian F.G. Schendera" <schendera@NIKOCITY.DE>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Christian F.G. Schendera" <schendera@NIKOCITY.DE>
Subject:      simultaneous computation of two functions of annual developments
              in percentages (first solved, second perhaps)
Comments: To: SASL <SAS-L@UGA.CC.UGA.EDU>
Content-Type: text/plain; charset="iso-8859-1"

Hi there!
Obviously I suffered from a heavy programmer's block. Over the weekend I
sent the following two questions. The first is solved for sure (about the
second I am a bit unsure):
>I have to compute two annual developments in percentages. Both computations
>have their peculiarities. As the second bases on the first analysis, it
>would be wonderful to code them  it in such a way that both questions could
>be analyzed simultaneously.
>1. Difference within (!) each YEARPAIR (see note below) in percentages (per
>BIRDTYPE and METHOD).
SOLUTION (very easy)
:            PDIFF = QJ2 - QJ1 ;
            PCTDIFF = (PDIFF/QJ1)*100 ;

Output looks like this:
  OBS    JAHRPAAR    JAHR1    JAHR2    VOGELART                 METHODE
QJ1     QJ2    PDIFF     PCTDIFF
             1     92/93       1992     1993    Amsel                    L
16.5    21.5      5.0      30.303
             2     92/93       1992     1993    Bachstelze               L
18.5    12.0     -6.5     -35.135
             3     92/93       1992     1993    Baumpieper               L
>2. Now I want to standardize the obtained % differences. The start value is
>100%, then I apply the PCTDIFFS to the 100%. Result could look like this
>(PCTDEV= %).
>BIRDTYPE  METHOD                          92/93
>93/94.............
>blackbird             L                +124.24 (+24.24 to start value 100%)
>112.68 (124.24-9.30%).............
>finch                    P                   ................
Here I use the lag function (although I am not very sure if there is a more
elegant solution):
        data PCTVLF2 ;
         set PCTVLF1 ;
           by VOGELART METHODE ;
            a=lag(PCTDIFF) ;
             if first.METHODE then PCTDEV=100 ;
             else PCTDEV =(PCTDIFF/a*100)-100 ;
            put PCTDEV = ;
          run ;
Hints very welcome.
Chris