Subject:
|
Re: Rate the factions
|
Newsgroups:
|
lugnet.admin.general
|
Date:
|
Fri, 13 Sep 2002 17:35:46 GMT
|
Viewed:
|
1385 times
|
| |
|
|
In lugnet.admin.general, Bruce Schlickbernd writes:
> Yes, but are The Powers That Be willing to go to the trouble of programming
> that?
Question is more likely, is it easily feasible? Doing a more-or-less
standard select,sum,and average query on a database is pretty easy, and is
pretty optomized. Picking out the X high and X low votes and removing them
before taking the average is a bit more complicated. If done within code
rather than query, it wouldn't be too tough-- but the other way 'round would
make it quite a difficult change. I'm confident that the result would be
more accurate, though.
> > I actually tried this at the time (a few months after 6067 was re-released)
> > with set rankings, with different values of N-- and oddly enough, 6067 was
> > ranked #1 more than with any other set for any value of N tried (I tried
> > something like 10 values of N? I forget offhand). And (at the time), 6067
> > had been ranked something like #20 or something, when it should have been at
> > least within the top 3, if not #1.
Went back and checked my old results-- using values of N 3 through 40 (and
without using the method), there were 19 sets that made it to the various
top 10's. (I tested on all 4,624 sets in the DB at the time, which was about
Dec. 11th, 2001). Anyway, out of these 39 methods, 6067 ranked #1 26 times.
6399 (the set that showed up as #1 at the time) ranked #1 only *once*, and
that was (of course), without using the method. 8480 ranked #1 6 times, and
6285 ranked #1 another 6 times. Note that both 6285 and 6067 had "0" votes
at the time, and were not even in the reported top 10, when they should have
been in 1st, 2nd, or 3rd.
Oh, and note also that in removing the topmost and bottommost votes, the
automatic "50" vote was never removed in my tests...
DaveE
|
|
Message is in Reply To:
 | | Re: Rate the factions
|
| (...) Ahhhh, I knew there was something I forgot. I looked at all the set ratings, and noticed some weinie had gone through and rated a bunch of very good sets as 0, in a fit of ratings trashing. I was curious as to where things would fall out if (...) (22 years ago, 13-Sep-02, to lugnet.admin.general)
|
2 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
