BP Stats Machine

[jwinslow]

Hey, I'm not sure where to post this question... but grad alluded to a bp stats machine being worked on... A long time ago Clarity wanted to have the best stats database ever which would not only have every normal type of stat... but also let you see how this running back did against these teams... or see how tOSU fared against a particular conference or school

Anyway, I was wondering who out there was working on this and whether it was still possible to pull it off... or if it might be better to go with something smaller in the meantime.

For references sake, I tried pulling up stats and found that the official ohio state buckeyes stats are really incomplete. These aren't unbelievably in depth either, but they are pretty useful for ranking the players based on stats:

http://msn.foxsports.com/cfb/teamStats?categoryId=86061

 

[Ratboy18]

I have to create a searchable online database for one of my classes, maybe I could work on this OSU project for class -- where are the stats currently residing?

 

[osugrad21]

Clarity is the man to answer this but as far as I know the site programmer is doing it in pieces as $$ is available

 

[OilerBuck]

The main thread demonstrating the database structure is here.

We seemed to be pretty close to the database being completed last July, so I cannot imagine much more of the work being there.

I anticipate the most help will be needed in the stats. It is going to take forever to compile all the stats that need to be gathered and then to break them out into tables and using the appropriate keys to make sure it all relates.

Since very few people on the site are probably familiar with database structuring, this could be difficult for data entry. I have to believe we will need as many people as possible to look up this data, but we would have to be able to count on people to make sure the stats are 100% accurate or the database is rendered worthless.

I think a data workbook should be made and sent to those chosen to look up this information. I forsee these people each being assigned a game at a time (starting with the 1889? season), they will need to research every possible detail on these games, complete the workbook and report their findings back (with unknown fields) to one of several project managers who would try their best to research the unknowns and would enter the information into the database once all resources are exhausted.

Once this is finished, the front-end will be mainly PHP/mySQL programming. I don't think that will be very difficult, as Clarity has proven he has resources that are very capable of customizing this site.

I'm not claiming any in-depth knowledge of this project...but I have to assume the statistical research is the most daunting task.

 

[WestEnd]

edit: holy shit! after reading that other thread, I'm blown away by what is envisioned. I was just thinking that you would do something simple, like parsing the game box scores into a database for a state engine. Clarity has alot more than a stats engine layed out, he's got the infinite buckeye information nuclear power plant of stats engines proposed.

--------------------------------------

I don't know anything about online programming, but ...

Really, I'll bet the best way to do it would be to copy the box scores from every game to a webpage on buckeyeplanet that would have a custom/uniform setup. that way, if you were to get the data off of cbs.sportsline or whatever, and they changed their format, it wouldn't screw it all up.

Really, for discussions sake, (and especially for the sake of giving Ratboy some ideas if he is actually going to do it) the interesting thing would be deciding what database fields we could add to each game in addittion to the box scores that you'll get from whatever website has the best game-by-game boxscores. Off the top of my head (and as have already been mentioned)...

opponant
home or away (or neutral)
day or night
grass or turf
nationally televised or not
injuries(? this might require a little bit of searching for the past games, but if you came up with a standard method of entry, it would be easier when future box scores were entered into the database)
class (ie it would be cool to pull up the cummulative stats for an entire recruiting class ... something like comparing the number of tackles the 02 class of LBs made vs. the 04 )

just my two cents. It sounds like a lot of work, but you'd definately get an A+ from me if I were youre prof.

 

[osugrad21]

OB, I'm not 100% sure, but I believe that is what the programmer is being paid to do...

I believe he is doing everything from construction to stat implementation...

At least that's what I have gathered

 

[OilerBuck]

OB, I'm not 100% sure, but I believe that is what the programmer is being paid to do...

I believe he is doing everything from construction to stat implementation...

At least that's what I have gathered

If that is the case, he's got a heck of a project on his hands. The database design and the site programming would be hard enough, but I guarantee the work is about 90-95% research and data entry and 5-10% technical.

Like Grad said, Clarity would know best...but I have to believe the stats engine is in capable hands.