Catapult -- GPS -- GT Students

[beerbuzz]

https://www.ajc.com/sports/college/...y-improve-performance/fGG7xiCxrLwKzyatt26dgM/

I've wondered ever since I saw the catapult gadgets.......wouldn't it be cool if a professor led students in a project/study to mine the data and come up with new ways to use it. It has to be coming. You can get a degree at GT in analytics. https://analytics.gatech.edu/ The IE department has even gotten involved with Big Data.

Can't believe GT didn't start this a long time ago (4-5 units.....pfffttt).

I'm sure Catapult shows how they think it can/should be used, but if GT students are worth their salt, they should be able to come up with proprietary uses that blow Catapult's doors off. I would have killed for a project like this back in the day.

 

[MACHETE]

UC Berkeley offers degrees in Big Data - “Division of Data Science and Information”.

 

[00Burdell]

Maybe someone understands how a GPS system can take one thousand intervals per second and capture the incremental motion but I don't. Is there a local reference point?

 

[Flywheel]

I'd be using machine learning to produce for coaches an assessment of the most likely plays to succeed, based on sheer statistics, for each play call.

3rd and 2 on the 35, 4th quarter and down by 3, then add height, weight, speed and "hands" rating of receivers, similar stats for other positions, and success rates for various types of plays for both teams. Kick out a ranked and scored priority list of possible play calls based on chance of success and how many times you've run it already to help coaches make decisions and pipe it to their headsets.

There's nothing in the way of datamining college football's thousands upon thousands of plays each year to do this. I could do it in simplified form in Access with SQL if I had enough time. The real challenge would be updating it as you go faster than the play clock with info from the previous play and player substitutions. That's where scripts and AI would be needed most.

 

[00Burdell]

I'd be using machine learning to produce for coaches an assessment of the most likely plays to succeed, based on sheer statistics, for each play call.

Now that's an idea that gets the gears turning...

 

[User 10337]

I'd be using machine learning to produce for coaches an assessment of the most likely plays to succeed, based on sheer statistics, for each play call.

3rd and 2 on the 35, 4th quarter and down by 3, then add height, weight, speed and "hands" rating of receivers, similar stats for other positions, and success rates for various types of plays for both teams. Kick out a ranked and scored priority list of possible play calls based on chance of success and how many times you've run it already to help coaches make decisions and pipe it to their headsets.

There's nothing in the way of datamining college football's thousands upon thousands of plays each year to do this. I could do it in simplified form in Access with SQL if I had enough time. The real challenge would be updating it as you go faster than the play clock with info from the previous play and player substitutions. That's where scripts and AI would be needed most.

Then we could have teams of people looking for other teams’ Catapult information to input into our system.