stephenbrooks.orgForumMuon1GeneralSome stats observations
Username: Password:
Search site:
Subscribe to thread via RSS
MereMortal
2003-11-01 19:33:11
Since I'm having to redo the retrieval script for this project for my site, here are a few musings/questions/gripes about the stats:

1) Is [DPC]NGS really a team?  Its only member is included in both the [DPC]NGS & the [DPC] teams.  If I was a member of a team ranked lower than [DPC]NGS, I'd have to be asking myself, "What's up with that?"

2) Are the team identifiers listed in teamids.txt required identifiers or just suggested identifiers?  If I was to parse rawstats.txt based on the identifiers listed in that file, I'd miss a good many of the people that are listed on the team stats page.  Some of the problem is with case sensitivity, but much of it seems to be that some folks just don't use these identifiers in their names, yet they are still listed on the teams stats page.  For example, most of rechenkraft.net uses the website name, not the team identifier.

3) Related to (2) is something I ran across when testing my script.  Strangely enough, it also pertains to rechenkraft.net, although it could be more widespread than my chance search pattern.  I happened to do my regular expression search on "rechenkraft|Rechenkraft" and noticed that the number of total particles was higher than that listed on the teams stats page.  It turns out one of their members has a "team identifier" as "www.rechenkraft.de" and is not listed on the teams page.  Without stricter enforcement of the team identifiers, minor differences in identifier names are going to be missed.  If anyone from rechenkraft is reading this, you might want to tell your teammate Zero that his/her name is exactly how much credit he/she is receiving for your team.


Ok, I'm done.  Razz Smile Now I have to go figure out how to make the Climate Prediction guys list more than the top 200 teams.  My work is never done!  Roll Eyes Wink

I hate stats more than life itself.
ZeonX[OCAU]
2003-11-02 02:24:09
Yeah the identifiers and what people use were very different also some use them at the end some at the front and 1 i found i the middle.

So when i made my stats i came across this problem but with a few similarities found in them i made my grabber smart and it made it so it knew what team and removed any part of the team identifier even if they added brackets or didn't have them and so on, took a while but works a dream now.  My stats aren't going because of the new 15cm stats actully made mine go whobbly on me and there are enough stats around without mine to add.
[DPC]Stephan202
2003-11-02 03:28:34
As for the [DPC]NGS thing: This was originally an idea of Stephen.  I think it's a very good idea, because this allowes us at DPC to easily create subteam stats.  Atm it's only one user using this construction, but in the future there may be many more.

---
Dutch Power Cow.
MOOH!
Herb[Romulus2]
2003-11-02 04:22:11
quote:
I hate stats more than life itself. 
Dave is a liar Big Grin Razz

-------------------------------
I'd say more, but I can't reach the keyboard from the floor.
Stephen Brooks
2003-11-02 04:22:31
The team identification is done by case-insensitive substring detection, stristr in C/PHP-speak.  REGEXP is nasty for doing case-insensitive stuff unless you have a special command to make your match case-insensitive.

It doesn't make any sense: that's why they call it "virtual"
[DPC]Stephan202
2003-11-02 04:37:44
quote:
Originally posted by Stephen Brooks:
The team identification is done by case-insensitive substring detection, stristr in C/PHP-speak.  REGEXP is nasty for doing case-insensitive stuff unless you have a special command to make your match case-insensitive.

It doesn't make any sense: that's why they call it "virtual"


That's easy, just add /i. Still, stristr is much faster (afaik), because it lacks all the fancy stuff.  Good to hear I used the right function in my stats code.

---
Dutch Power Cow.
MOOH!
MereMortal
2003-11-02 08:50:06
quote:
Originally posted by [DPC]Stephan202:
As for the [DPC]NGS thing: This was originally an idea of Stephen.  I think it's a very good idea, because this allowes us at DPC to easily create subteam stats.  Atm it's only one user using this construction, but in the future there may be many more.



Well, I guess for my own nefarious purposes, listing subteams in is anti-helpful.  I have decided that I'm not going to include them when determining rankings, as I think it is unfair to other teams.  Of course, now I'll have to explain this to all the e-mails that I get complaining that the rankings don't match the official rankings.

As for the case insensitive stuff, that's just my bone-headed oversight---that's what happens when you try to code, referee a paper, and watch football at the same time.  Roll Eyes Speed isn't a great concern for this code, as any speedup by using stristr is a tiny fraction of the overall retrieval time when you are dealing with 28 projects, and it's only run once per day.

Thanks for your comments!

-----
I hate stats more than life itself.
magnav0x
2003-11-02 17:48:40
quote:
Ok, I'm done.  Now I have to go figure out how to make the Climate Prediction guys list more than the top 200 teams.  My work is never done! 


Just finished my Climate Prediction stats over at http://stats.dbestern.net/climateprediction

I also finished my Muon statistics, just waiting a day or two to make sure everything is solid before releasing the link.
[OCAU] badger
2003-11-02 19:20:45
would it be possible to have links to these stats pages?

www.BadgerMotorsport.tk Proudly sponsored by GRX-Computers
magnav0x
2003-11-02 19:56:47
erm...I see no stats at those pages....
Stephen Brooks
2003-11-03 01:04:17
quote:
Originally posted by [DPC]Stephan202:
That's easy, just add /i.
Yeah, or even if your regexp command doesn't have this feature you can use strlwr/strtolower to normalise before matching.  But seriously you'd be surprised what some people get up to: I've seen serious computer scientists discussing how annoying it is that they have to use (T|t)(H|h)(I|i)(S|s) or some such thing to do insensitive matching!  Big Grin

It doesn't make any sense: that's why they call it "virtual"
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel charactermstdn.io

Site has had 17091333 accesses.