Re: st: RE: Programming Problem

[Antoine Terracol <Antoine.Terracol@univ-paris1.fr>]

Guido,

Here's a quick-and-dirty solution that seems to work. it uses local 
macros and scalars, so it's pretty fast. I guess you'll find a way to 
pass the number of votes from the dataset to scalars.
however, the sequence of dividers here is "1,2,3,4,...,m"

Hope this helps,
Antoine.
----------------------------------------------------------
clear

/*fill in the number of seats and the number of parties*/
local totalseat=6
local nbparties=4


/*fill in the number of votes*/
scalar nbvp1=41000
scalar nbvp2=29000
scalar nbvp3=17000
scalar nbvp4=13000


local i=1
while `i'<=`nbparties' {
           local nbs`i'=0
           scalar nbv`i'=scalar(nbvp`i')  /*scalars to be divided 
during the loop*/
           local i=`i'+1
           }


local string "scalar(nbv1)"
local i=2
while `i'<=`nbparties' {
           local string "`string', scalar(nbv`i')" /*builds the list of 
votes to be compared*/
           local i=`i'+1
           }

local i=1
while `i'<=`totalseat' {
           local j=1
           local done=0
           while `j'<=`nbparties' {
                           if scalar(nbv`j')==max(`string') & `done'==0 {
                                                           local 
nbs`j'=`nbs`j''+1   /*increment the number of seats*/
                                                           local done=1
                                                           scalar 
nbv`j'=scalar(nbvp`j')/`nbs`j''  /*divide total votes by number of seats+1*/
                                                           }

                           local j=`j'+1
                           }
           local i=`i'+1
           }
                          
local i=1
while `i'<=`nbparties' {
           di "Party `i' gets `nbs`i'' seats"
           local i=`i'+1
           }

------------------------------------------------------
Guido Tiemann wrote:


> At 11:17 19.07.2002 -0400, you wrote:
>
> > I'm not entirely clear on the algorithm here.  Is it:
> >
> > 1. Assign a seat to current highest vote getter.
> > 2. Divide that party's votes by (1+number of seats they've been awarded
> > so far).
> > 3. Return to step 1, until all seats are assigned.
>
>
> Yep, you are quite right, Nick. Sorry, if I did not clarify this 
> point. There is only one qualification, I would like to add to my 
> former posting: The sequence of divisors might also be a different 
> one, for instance, 1 - 3 - 5 (or even 1.4 - 3 - 5) for the method of 
> (modified) Sainte Lague. In addition, I would like to preserve the 
> initial votes that will be needed for further calculations, and the 
> data set is rather huge (about 200 meg). As a consequence, my "brutal" 
> slope + dummy approach took really long ...
>
> Guido Tiemann
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/