# Thread: F@H Statistics

1. ## F@H Statistics

My friend and I have begun work on a folding at home statistics site.

I'm gonna break F@H down for you all real quick.
Every 3 hours a Team submits more points and "Units" to their total points and Unit count. (The unit is where the points come from). So every 3 hours Point and Unit totals grow for each unique team.
A team consists of Users who submits the points and units that a team unites under one ID.
That is to say, User 1 and User 2 submit 4 units each to Team 1 and then Team 1 logs 8 Units.

If you want more details as to what F@H is: http://folding.stanford.edu/

There are a few stat sites already in existence and they can be found here:

Folding @ Home Teams Overall Rank - EXTREME Overclocking Folding @ Home Stats
Kakao Stats - Teams Ranking

Most of the stat sites that I know track very basic statistics such as average daily submission amount per team/user. Or things such as how much an individual is expected to have in a year.

It's all pretty basic, and you can see more on the sites I provided.

So what I'm hoping you all can provide me with is really cool ideas on some stuff my friend and I can do with these stats. I have taken only an intro. stat class which has yielded a few ideas we might use but I haven't, nor has my friend, taken any advanced stat classes. Albiet, my friend, who happens to be doing all the back-end "math coding" is a veritable math whiz. If you can just name some advanced stat topics and what it would yield I'm sure he could look it up and code some stuff to go on the site.

So please don't hold back your recommendations.

2. There's a chance I'll look at this at some point and come up with suggestions. A point of interest, for experts who visit this board, is that this Stanford protein folding stuff actually uses some pretty hardcore probabilistic models. Persi Diaconis gave a lecture on Markov Chains at my university this fall and much of the work he did was motivated by this project (essentially, for computational reasons the researchers wanted to see if the protein folding could be modeled well by a Markov Chain within a Bayesian framework). So, really interesting stuff from a statistics/probability theory point of view is here as well.

3. Originally Posted by theodds
There's a chance I'll look at this at some point and come up with suggestions. A point of interest, for experts who visit this board, is that this Stanford protein folding stuff actually uses some pretty hardcore probabilistic models. Persi Diaconis gave a lecture on Markov Chains at my university this fall and much of the work he did was motivated by this project (essentially, for computational reasons the researchers wanted to see if the protein folding could be modeled well by a Markov Chain within a Bayesian framework). So, really interesting stuff from a statistics/probability theory point of view is here as well.

Well, just so you know, I'm only looking for statistics suggestions regarding the points that users get for submitting WU's. I'm not looking for anything related to the research specific statistics.