1. correlation coefficient program
- Posted by Jerry Story <story.jerry at ?mail.co?>
Sep 04, 2007
-
Last edited Sep 05, 2007
Some time ago there was some discussion about correlation coefficient.
http://www.openeuphoria.org/cgi-bin/esearch.exu?fromMonth=6&fromYear=2&toMonth=9&toYear=C&postedBy=Craig&keywords=%22correlation+coefficient%22
I swiped a huge bunch of data files and made a wxEuphoria-based program using
Rob Craig's code.
http://www3.telus.net/~jtstory/programs-unfinished/correlations.zip
Is there any use for this type of program?
The data files I used
http://lib.stat.cmu.edu/DASL/allsubjects.html
don't seem to produce very useful results.
Is there any computer-readable data to make actually useful
correlations out of?
2. Re: correlation coefficient program
Jerry Story wrote:
>
> Some time ago there was some discussion about correlation coefficient.
> <a
> href="http://www.openeuphoria.org/cgi-bin/esearch.exu?fromMonth=6&fromYear=2&toMonth=9&toYear=C&postedBy=Craig&keywords=%22correlation+coefficient%22">http://www.openeuphoria.org/cgi-bin/esearch.exu?fromMonth=6&fromYear=2&toMonth=9&toYear=C&postedBy=Craig&keywords=%22correlation+coefficient%22</a>
>
> I swiped a huge bunch of data files and made a wxEuphoria-based program using
> Rob Craig's code.
> <a
> href="http://www3.telus.net/~jtstory/programs-unfinished/correlations.zip">http://www3.telus.net/~jtstory/programs-unfinished/correlations.zip</a>
>
> Is there any use for this type of program?
>
> The data files I used
> <a
> href="http://lib.stat.cmu.edu/DASL/allsubjects.html">http://lib.stat.cmu.edu/DASL/allsubjects.html</a>
> don't seem to produce very useful results.
> Is there any computer-readable data to make actually useful
> correlations out of?
Most data series related with economics have high autocorrelation coefficients
for small lags, because economic systems have a large inertia: the value you
measure at some point will be quite strongly related to the one(s) you measured
shortly before. Then you'll have useful stuff.
CChris
3. Re: correlation coefficient program
CChris wrote:
>
> Jerry Story wrote:
> >
> > Some time ago there was some discussion about correlation coefficient.
> > <a
> > href="http://www.openeuphoria.org/cgi-bin/esearch.exu?fromMonth=6&fromYear=2&toMonth=9&toYear=C&postedBy=Craig&keywords=%22correlation+coefficient%22">http://www.openeuphoria.org/cgi-bin/esearch.exu?fromMonth=6&fromYear=2&toMonth=9&toYear=C&postedBy=Craig&keywords=%22correlation+coefficient%22</a>
> >
> > I swiped a huge bunch of data files and made a wxEuphoria-based program
> > using
> > Rob Craig's code.
> > <a
> > href="http://www3.telus.net/~jtstory/programs-unfinished/correlations.zip">http://www3.telus.net/~jtstory/programs-unfinished/correlations.zip</a>
> >
> > Is there any use for this type of program?
> >
> > The data files I used
> > <a
> > href="http://lib.stat.cmu.edu/DASL/allsubjects.html">http://lib.stat.cmu.edu/DASL/allsubjects.html</a>
> > don't seem to produce very useful results.
> > Is there any computer-readable data to make actually useful
> > correlations out of?
>
> Most data series related with economics have high autocorrelation coefficients
> for small lags, because economic systems have a large inertia: the value you
> measure at some point will be quite strongly related to the one(s) you
> measured
> shortly before. Then you'll have useful stuff.
>
> CChris
Correlation and autocorrelation are related concepts, but not the same.
Correlation refers to the relation between two parallel data series, for example
age and IQ, or country gross product and AIDS incidence, etc.
On the other hand, autocorrelation refers to the relation between a data series
and the same, displaced some places, which usually applies to data that is a
function of time.
A program to compute correlation may be usually applied also to compute
autocorrelation.
I haven't browsed the data you are referring to, but maybe you can try to
correlate different weather data: for example air humidity with rain next day,
etc.
Regards.
