FW: [plt-scheme] Statistics (V301.5 Speed Up)

From: Joe Marshall (jmarshall at alum.mit.edu)
Date: Sat Feb 11 17:09:03 EST 2006

The attached graph here is most definitely *not* gaussian.
I'd say Jim is off the hook.

On 2/11/06, Williams, M. Douglas <M.DOUGLAS.WILLIAMS at saic.com> wrote:
> Actually, this exercise is very much a "plug and chug" endeavor on my part.
> I am just curious - in a wholly unscientific way - about the speed up from
> V301 to V301.5 for a continuation entensive application.  In this case, the
> application domain is models developed using the simulation collection I
> developed.  It was meant as a thanks for the work that went into the
> run-time improvements and as I way that I could guage for myself some
> measure of the speed up.  I am quite satisfied with simple statistics on the
> results - the only variable between runs of the same model has been the
> version of the PLT Scheme system used.  Some of us were also curious about
> the shape of the distributions of run times - another example of which is
> attached here.
> If anyone would like to do some statistically meaningful ananysis on the
> data.  All of the code is available on PLaneT.
> Doug
> -----Original Message-----
> From: Matt Jadud [mailto:mcj4 at kent.ac.uk]
> Sent: Sat 2/11/2006 1:15 AM
> To: Williams, M. Douglas
> Cc: plt-scheme at list.cs.brown.edu
> Subject: Re: FW: [plt-scheme] Statistics (V301.5 Speed Up)
> Statistics is not a "plug-n-chug" endeavor. One of my favorite resources
> to date on exploratory data analysis comes from NIST:
> http://www.itl.nist.gov/div898/handbook/eda/eda.htm
> I also liked Trisha Greenhalgh's "How to read a paper: The basics of
> evidence-based medicine", a book (surprise!) on interpreting reports and
> results in the area of evidence-based medicine. The full text seems to
> be available here:
> http://bmj.bmjjournals.com/collections/read.shtml
> and the two articles "Statistics for the non-statistician" and
> "Statistics for the non-statistician II" are good; the remainder of the
> book requires some interpretation to apply out of context, but it is
> still a good resource while performing EDA on unknown datasets.
> http://bmj.bmjjournals.com/cgi/content/full/315/7104/364
> http://bmj.bmjjournals.com/cgi/content/full/315/7105/422
> Sadly, I can't join further in the First International Scheme Contest
> for One Statistical Beer, as I need to take off for parts uknown for the
> day.
> *sigh*
> No free beer for me.
> M
> Williams, M. Douglas wrote:
> > I ran 1000 runs and made a histogram of the results.  Unfortunately, there
> > is one value out at 16353 while the others were between about 6000 and
> 8000.
> > Which made the histogram difficult to interpret.  I have attached the code
> > and the resulting histogram.
> _________________________________________________
>  For list-related administrative tasks:
>  http://list.cs.brown.edu/mailman/listinfo/plt-scheme


Posted on the users mailing list.