Robert,<br><br>If you don't pass the set.seed, R will use a random seed already I believe. The reason I use it with the same seed is that I had a need to be able to run the query multiple times on the same data and be sure I would get the same results. I've had great success using this approach with a number of different R functions.
<br><br>I'm glad to see that you are getting some use out of it too.<br><br>David<br><br><div><span class="gmail_quote">On 1/25/07, <b class="gmail_sendername">Robert Burgholzer</b> <<a href="mailto:rburghol@chesapeakebay.net">
rburghol@chesapeakebay.net</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">David/others,<br>I have finally revisited this thread, having forgotten all about it, and
<br>missing the finale. I wish I had realized you solved my problem already<br>(with your array_accum), as I came up with a less elegant solution --<br>after hours of struggle :).<br><br>Anyhow, I have modified your "kmeans" function slightly to make it a bit
<br>more robust (I think), allowing it to use decimals instead of integers,<br>and allowing you to pass in the seed value yourself (assuming that some<br>utility exists in being able to supply a random, rather than static,
<br>seed). That said, I don't really know if kmeans is supposed to act on<br>non-integer values, but it seems to behave OK.<br><br>CREATE OR REPLACE FUNCTION kmeans(double precision[], int4, int4)<br> RETURNS double precision[] AS'
<br>set.seed(arg3)<br>km=kmeans(sort(arg1),arg2)<br>sort(unlist(tapply(sort(arg1),factor(match(km$cluster,order(km$centers))<br>),range)))<br>' LANGUAGE 'plr' VOLATILE STRICT;<br><br>I have posted this code, as well as an implementation of the "quantile"
<br>function (now using your more robust array_accum implementation) at:<br><br><a href="http://soulswimmer.dynalias.net/db/R/r_functions.01.sql">http://soulswimmer.dynalias.net/db/R/r_functions.01.sql</a><br><br>Comments, suggestions, and other R function implementations are most
<br>welcome.<br><br>r.b.<br><br><br>-----Original Message-----<br>From: <a href="mailto:postgis-users-bounces@postgis.refractions.net">postgis-users-bounces@postgis.refractions.net</a><br>[mailto:<a href="mailto:postgis-users-bounces@postgis.refractions.net">
postgis-users-bounces@postgis.refractions.net</a>] On Behalf Of<br>David Bitner<br>Sent: Thursday, March 02, 2006 12:45 PM<br>To: PostGIS Users Discussion<br>Subject: Re: [postgis-users] quantiles, quartiles, or jenks natural
<br><br>By the way, the set.seed call is so I get the same results with<br>subsequent calls on the dataset as I make one call with PHP and<br>PostgreSQL to the database to create my legend with the class<br>intervals and another to divvy up my dataset into predefined class
<br>styles in my mapfile with MapServer and I need subsequent calls to<br>come up with the same results.<br><br>On 3/2/06, David Bitner <<a href="mailto:osgis.lists@gmail.com">osgis.lists@gmail.com</a>> wrote:<br>> The main function that I made is kmeans which takes an array of the
<br>> values that you want to classify and the number of classes that you<br>> want and spits out an array of the break points for the data.<br>><br>> For example:<br>> select<br>kmeans(array[1,1,1,1,4,3,5,6,45,3,5,7,8,6,4,3,2,1,32,6,7,5,6,7,8],4)
<br>> returns<br>> {1,2,3,4,5,8,32,45}<br>> which can be interpreted as use these classes:<br>> 1-2,3-4,5-8,32-45<br>> which extending to have no gaps would be the same as either<br>> 1-2,3-4,5-31,32-45 or
<br>> 1-2,3-4,5-8,9-45<br>><br>> I generally just call this using an array_accum aggregate like this:<br>> select kmeans(array_accum(myintegercol),4) from mytable<br>><br>> As I said before, I kept getting some parse errors that I haven't had
<br>> time to look into when I tried writing the function to multiple lines,<br>> so the function is all one line.<br>><br>> CREATE OR REPLACE FUNCTION kmeans(_int8, int4)<br>> RETURNS _int8 AS<br>><br>
'set.seed(2007);km=kmeans(sort(arg1),arg2);sort(unlist(tapply(sort(arg1)<br>,factor(match(km$cluster<br>> ,order(km$centers))),range)))'<br>> LANGUAGE 'plr' VOLATILE STRICT;<br>><br>> CREATE AGGREGATE array_accum(
<br>> BASETYPE=anyelement,<br>> SFUNC=array_append,<br>> STYPE=anyarray,<br>> INITCOND='{}'<br>> );<br>><br>><br>><br>> On 3/2/06, Stephen Woodbridge <<a href="mailto:woodbri@swoodbridge.com">
woodbri@swoodbridge.com</a>> wrote:<br>> > David,<br>> ><br>> > Please post it to the listserv, I would be interested also. I have<br>yet<br>> > to jump into PL/R but it is on my list to do.<br>
> ><br>> > Thanks,<br>> > -Steve<br>> ><br>> > David Bitner wrote:<br>> > > I ended up jumping into the PL/R world and just created an<br>aggregate<br>> > > wrapper around kmeans to get my class values. They ended up being
<br>> > > very, very close (identical in some cases) to classifications that<br>had<br>> > > been done with Jenks Natural Breaks. If you want the same results<br>> > > every time you run a classification on the same data, you need to
<br>set<br>> > > the same seed value for the random number generator before each<br>run.<br>> > ><br>> > > It's pretty basic and my code is ugly due to some R parser errors<br>that<br>> > > I could only get passed by throwing all the code on one line with
<br>no<br>> > > spaces (hey it worked and I didn't have time to look into the<br>parser<br>> > > error), but I can throw the code up if anyone would like.<br>> > ><br>> > > On 3/2/06, Robert Burgholzer <
<a href="mailto:rburghol@chesapeakebay.net">rburghol@chesapeakebay.net</a>> wrote:<br>> > ><br>> > >>OK,<br>> > >>I'm coming into this late, but I am a user of PL/R and PostGIS,<br>
and<br>> > >>would appreciate any progress on developing some classification<br>routines<br>> > >>to be posted to this lists, or I would be interested in being<br>notified<br>> > >>offline.
<br>> > >><br>> > >>Thanks!<br>> > >><br>> > >>r.b.<br>> > >><br>> > >>-----Original Message-----<br>> > >>From: <a href="mailto:postgis-users-bounces@postgis.refractions.net">
postgis-users-bounces@postgis.refractions.net</a><br>> > >>[mailto:<a href="mailto:postgis-users-bounces@postgis.refractions.net">postgis-users-bounces@postgis.refractions.net</a>] On Behalf<br>Of Amit<br>> > >>Kulkarni
<br>> > >>Sent: Wednesday, March 01, 2006 1:20 PM<br>> > >>To: <a href="mailto:postgis-users@postgis.refractions.net">postgis-users@postgis.refractions.net</a><br>> > >>Subject: Re: [postgis-users] quantiles, quartiles, or jenks
<br>natural<br>> > >><br>> > >>Sorry, I have been catching up on the past few months emails. I<br>just<br>> > >>want to add that I read that quantiles and minimum boundary error<br>are<br>
> > >>better than jenks. Also minimum boundary error takes into account<br>the<br>> > >>underlying topology.<br>> > >><br>> > >>The two being better are mentioned in<br>> > >>
<br>> > >>Brewer, Cynthia A. & Pickle, Linda (2002) Evaluation of Methods<br>for<br>> > >>Classifying Epidemiological Data on Choropleth Maps in Series.<br>> > >>Annals of the Association of American Geographers 92 (4), 662-681
<br>> > >><br>> > >>And the minimum boundary algorithm is supposedly mentioned in<br>> > >><br>> > >>Cromley, E. K. , and R. G. Cromley. 1996. An analysis of<br>alternative<br>
> > >>classification schemes for medical atlas mapping. European<br>Journal<br>> > >>of Cancer 32A (9): 1551 -- 59.<br>> > >><br>> > >>Cromley, R. G. , and R. D. Mrozinski. 1999. The classification of
<br>> > >>ordinal data for choropleth mapping. The Cartographic Journal 36<br>(2):<br>> > >>101 -- 9.<br>> > >><br>> > >>HTH,<br>> > >>amit<br>> > >>
<br>> > >><br>> > >>Date: Tue, 14 Feb 2006 12:38:39 -0800<br>> > >>From: Paul Ramsey <<a href="mailto:pramsey@refractions.net">pramsey@refractions.net</a>><br>> > >><br>
> > >>I did some in PHP, but the algorithms are relatively braindead,<br>the<br>> > >>quantile stuff in particular. Jenks I did some research on but<br>never<br>> > >>really found a definitive description of the process. Some of the
<br>> > >>descriptions ended up sounding like a k-means clustering idea,<br>which<br>> > >>is not cheap!<br>> > >><br>> > >>P.<br>> > >><br>> > >>__________________________________________________
<br>> > >>Do You Yahoo!?<br>> > >>Tired of spam? Yahoo! Mail has the best spam protection around<br>> > >><a href="http://mail.yahoo.com">http://mail.yahoo.com</a><br>> > >>_______________________________________________
<br>> > >>postgis-users mailing list<br>> > >><a href="mailto:postgis-users@postgis.refractions.net">postgis-users@postgis.refractions.net</a><br>> > >><a href="http://postgis.refractions.net/mailman/listinfo/postgis-users">
http://postgis.refractions.net/mailman/listinfo/postgis-users</a><br>> > >>_______________________________________________<br>> > >>postgis-users mailing list<br>> > >><a href="mailto:postgis-users@postgis.refractions.net">
postgis-users@postgis.refractions.net</a><br>> > >><a href="http://postgis.refractions.net/mailman/listinfo/postgis-users">http://postgis.refractions.net/mailman/listinfo/postgis-users</a><br>> > >>
<br>> > ><br>> > > _______________________________________________<br>> > > postgis-users mailing list<br>> > > <a href="mailto:postgis-users@postgis.refractions.net">postgis-users@postgis.refractions.net
</a><br>> > > <a href="http://postgis.refractions.net/mailman/listinfo/postgis-users">http://postgis.refractions.net/mailman/listinfo/postgis-users</a><br>> > ><br>> ><br>> > _______________________________________________
<br>> > postgis-users mailing list<br>> > <a href="mailto:postgis-users@postgis.refractions.net">postgis-users@postgis.refractions.net</a><br>> > <a href="http://postgis.refractions.net/mailman/listinfo/postgis-users">
http://postgis.refractions.net/mailman/listinfo/postgis-users</a><br>> ><br>><br>_______________________________________________<br>postgis-users mailing list<br><a href="mailto:postgis-users@postgis.refractions.net">
postgis-users@postgis.refractions.net</a><br><a href="http://postgis.refractions.net/mailman/listinfo/postgis-users">http://postgis.refractions.net/mailman/listinfo/postgis-users</a><br>_______________________________________________
<br>postgis-users mailing list<br><a href="mailto:postgis-users@postgis.refractions.net">postgis-users@postgis.refractions.net</a><br><a href="http://postgis.refractions.net/mailman/listinfo/postgis-users">http://postgis.refractions.net/mailman/listinfo/postgis-users
</a><br></blockquote></div><br><br clear="all"><br>-- <br>************************************<br>David William Bitner