NEW: OPR Android App

[stingray27]

Hey ChiefDelphi,

I just posted my first Android App, OPR FIRST 2013, on the Google play market. The app's functionality includes:

• Live match results
• On-the-spot OPR calculations after match 30
• Match Predicting for each match based on OPR
• FIRST Team searching
• Award listings per regional
• Team listings per regional

This app will hopefully aid in pre-match scouting, selection process for elimination matches, or post-tournament analysis.

For more information and to download the app, please visit:
OPR FIRST 2013

Please leave any comments or bugs and I will try to fix them as soon as possible! Hope you enjoy!

[dodar]

Wow.

[cmrnpizzo14]

Originally Posted by dodar

Wow.

Agreed. If this works, this is incredible.

[MARS_James]

Originally Posted by dodar

Wow.

My thoughts exactly

[Ether]

What dataset did you use to validate your OPR calculations?

[stingray27]

Originally Posted by Ether

What dataset did you use to validate your OPR calculations?

I haven't... I was having trouble because I wasn't sure which OPRs integrated the adjusted score based on Co-op bridge balancing and which did not. So my data currently may be off slightly. I also am currently using the FRC-Spy xml feed to calculate the OPR. I really want to change that to the USFIRST data by parsing HTML because I know that data is accurate. I am open to specific methods of calculating it if you have any suggestions!

[Ether]

Originally Posted by stingray27

I am open to specific methods of calculating it if you have any suggestions!

I do have a suggestion.

Do not use matrix inversion. Instead, use Cholesky decomposition to factor the matrix. Read this thread, starting at post#36.

[Justin Montois]

Originally Posted by Ether

I do have a suggestion.

Do not use matrix inversion. Instead, use Cholesky decomposition to factor the matrix. Read this thread, starting at post#36.

Annndddd my brain exploded.


Definitely going to check out the app though! Good work.

[Ian Curtis]

Originally Posted by Ether

What dataset did you use to validate your OPR calculations?

Have you done any work to show that OPR on a partial event is useful? I looked at a regional in 2010 and found that that OPR really just wasn't worth looking at until Friday night, but would love to know if someone has done some more in depth work to show when it becomes useful, as the 1 regional I did is not exactly a big dataset.

[stingray27]

Originally Posted by Ian Curtis

Have you done any work to show that OPR on a partial event is useful? I looked at a regional in 2010 and found that that OPR really just wasn't worth looking at until Friday night, but would love to know if someone has done some more in depth work to show when it becomes useful, as the 1 regional I did is not exactly a big dataset.

Yes I have. That was a very interesting route I went when I looked at that information, because like you, I was curious to see what it would do. First, the reason I put a limit on when you can calculate OPR is because I cannot invert the matrix sooner than that match. Second, the data isn't quite representative until around 50% to 75% of the regional is complete. I have noticed that most of the OPRs are to the extremes when calculated early. As the regional progresses, the OPRs tend to settle into more valid ranges. By Friday night however, I am confident it will be fairly accurate and should do a good job in predicting the rankings.

[Andrew Schreiber]

Originally Posted by stingray27

Yes I have. That was a very interesting route I went when I looked at that information, because like you, I was curious to see what it would do. First, the reason I put a limit on when you can calculate OPR is because I cannot invert the matrix sooner than that match. Second, the data isn't quite representative until around 50% to 75% of the regional is complete. I have noticed that most of the OPRs are to the extremes when calculated early. As the regional progresses, the OPRs tend to settle into more valid ranges. By Friday night however, I am confident it will be fairly accurate and should do a good job in predicting the rankings.

Have you actually seen if OPR predicts better than a coin flip? I'd be curious at what point it exceeds the "pick random alliance" model or even the "pick alliance whose sum of rankings is lower" model.

[dtengineering]

Originally Posted by Ian Curtis

Have you done any work to show that OPR on a partial event is useful? .....

I'd say that the "predictions" for upcoming match results will be a good way to determine when the data starts to become significant. It would be great to have an analysis of what fraction of the predictions are correct... it would be an excellent way to show at what point the OPRs become an effective indicator of match outcome.

But OPRs and scouting reports are always useful, if only for entertainment's sake!

Jason

[Bongle]

Originally Posted by dtengineering

I'd say that the "predictions" for upcoming match results will be a good way to determine when the data starts to become significant. It would be great to have an analysis of what fraction of the predictions are correct... it would be an excellent way to show at what point the OPRs become an effective indicator of match outcome.

But OPRs and scouting reports are always useful, if only for entertainment's sake!

Jason

Older versions of OPRNet had a "self-check" mode for their predictor - if a regional was complete when someone asked for a prediction, it would tell you how good the predictions would have been if it had made them with the data available after each match. So it would say:
"After match 59, the predictions for matches 60-80 would have been 65% correct"
"After match 60, the predictions for matches 61-80 would have been 64% correct" (it varied a lot)

Generally what I found was that on friday night, saturday predictions based on the sum of OPRs for each alliance were 60-70% correct, which is obviously way better than a coin toss. This depends on the game though - OPR is better or worse depending on what the game is like. See the 1-hour mark of Karthik's presentation at champs last year for more details.

Post-event (aka once you know all the data) it was 80ish% correct, but that's a fairly worthless stat: once the event is over, anyone can get 100% of predictions correct!

[Racer26]

Am I the only one finding that this app straight up doesn't work? I couldn't pull anything for BAE all day, while FRC Spyder was working fine.

[engunneer]

Originally Posted by 1075guy

Am I the only one finding that this app straight up doesn't work? I couldn't pull anything for BAE all day, while FRC Spyder was working fine.

Bae results are working with todays update.

Thanks!