|
|
Re: File exchange downloads per author - an odd distribution
Posted:
Jan 2, 2009 10:23 AM
|
|
"Husam Aldahiyat" <numandina@gmail.com> wrote in message <gjl95l$3hp$1@fred.mathworks.com>...
> "Oliver Woodford" <o.j.woodford.98@cantab.net> wrote in message <gjl7dd$n0j$1@fred.mathworks.com>...
> > This code plots the distribution of number of downloads* per author for the top 100 authors**, as given by the stats on the FEX:
> >
> > s = urlread('http://www.mathworks.com/matlabcentral/fileexchange/authors');
> > s = regexp(s, '\d*(?=</td>[\n\c] *<td>\d*</td>[\n\c] *<td>\d* \w* \d*</td>[\n\c] *<td class="number">[\n\c] *<ul class="star-rating">)', 'match');
> > s = str2double(s);
> > hist(s, 100);
> >
> > The distribution seems very unnatural. In particular, there are a few authors with over 20,000 downloads, but none with 10,000 to 20,000 downloads. A bug, or coincidence?
> >
> > *Over the last 30 days
> > **In terms of number of downloads over the last 30 days
>
> Coincidence.
Actually, it was some people just screwing around. First, someone chose to play a joke on the infamous Marco by artificially inflating his downloads until he was #1. Then either those same individuals or others starting inflating the downloads of other high-ranked people to put them back ahead of Marco, dropping him to #8 (as of this post).
Maybe some of them wanted to make the point that the ranking can be easily fooled with since it is based on downloads?... or maybe they were just having fun.
Ken
|
|
