Chief Delphi

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)
-   General Forum (http://www.chiefdelphi.com/forums/forumdisplay.php?f=16)
-   -   Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting (http://www.chiefdelphi.com/forums/showthread.php?t=75272)

fordchrist675 16-03-2010 13:53

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
I have a question about Oregon for 2010. I was trying to find out opr and pm for teams at OR this year. and it failed. any ideas?

Bongle 16-03-2010 14:11

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Quote:

Originally Posted by fordchrist675 (Post 938065)
I have a question about Oregon for 2010. I was trying to find out opr and pm for teams at OR this year. and it failed. any ideas?

Mine failed from USFIRST but managed to use TBA as a backup.
Code:

Pos  Reg  Team    OPR    SAA      PM    RPI
  1  or  997    4.44  2.997  1.443  0.2869
  2  or  753  3.342  2.472  0.8707  0.2321
  3  or  3165  3.214  2.257  0.9566  0.2528
  4  or  1540  2.621  1.439  1.182  0.2585
  5  or  3192  2.537  2.571-0.03447  0.1906
  6  or  2910  2.168  0.6331  1.535  0.2883
  7  or  2865  2.045  2.172 -0.1265  0.1994
  8  or  948  1.979  2.207 -0.2276  0.2096
  9  or  2471  1.975    1.27  0.7048  0.2358
  10  or  1432  1.922  1.494  0.4277  0.213
  11  or  368  1.915  2.293 -0.3781  0.2302
  12  or    8  1.825  1.703  0.1226  0.271
  13  or  3145  1.814  1.986 -0.1718  0.2204
  14  or  957  1.791  2.599 -0.8078  0.1842
  15  or  847  1.709  0.6439  1.066  0.2785
  16  or  2811  1.701  0.1783  1.522  0.3023
  17  or  3223  1.692  1.6860.005676  0.2277
  18  or  2046  1.685  1.653 0.03189  0.1667
  19  or  1983  1.631  0.7979  0.8326  0.2701
  20  or  2557  1.466  0.9125  0.5539  0.3004
  21  or  488  1.444  1.262  0.1814  0.2307
  22  or  2733  1.373    1.67 -0.2969  0.2069
  23  or  2130    1.37  1.268  0.1024  0.2588
  24  or  3024  1.366  2.677  -1.312  0.1901
  25  or  1595    1.33  2.216 -0.8864  0.2402
  26  or  949  1.251  2.401  -1.15  0.2016
  27  or  1823  1.191  1.527  -0.336  0.2294
  28  or  2542  1.172  1.343 -0.1711  0.2331
  29  or  1700  1.089  1.579 -0.4898  0.2202
  30  or  3131  1.004  1.263  -0.259  0.1614
  31  or  3210  0.938  0.4483  0.4896  0.2581
  32  or  2374  0.914  0.3753  0.5387  0.255
  33  or  2951  0.8851  2.232  -1.347  0.1808
  34  or  1515  0.8275  0.6605  0.167  0.177
  35  or  2147  0.7307    1.34 -0.6097  0.2248
  36  or  2002  0.6813  0.3311  0.3502  0.2314
  37  or  1510  0.6691  0.7407-0.07164  0.2381
  38  or  2605  0.6604  1.997  -1.337  0.2139
  39  or  1425  0.5979  1.283  -0.685  0.2014
  40  or  2517  0.5943  0.2731  0.3212  0.2188
  41  or  3070  0.5522  0.4322    0.12  0.2002
  42  or  2898  0.5402  0.4317  0.1084  0.2305
  43  or  2522  0.5236  0.7541 -0.2305  0.199
  44  or  2192  0.5206  0.8281 -0.3075  0.2332
  45  or  955  0.5094  0.50780.001674  0.2162
  46  or  3311  0.4487 -0.1401  0.5888  0.1892
  47  or  2922  0.4001    1.29 -0.8898  0.221
  48  or  2411  0.3854  0.5043  -0.119  0.2797
  49  or  3188  0.3425  0.7973 -0.4548  0.2466
  50  or  2915  0.3389 -0.1069  0.4458  0.2668
  51  or  2550  0.2831  0.6419 -0.3587  0.2233
  52  or  3213  0.2749  0.5467 -0.2718  0.2069
  53  or  1571  0.2742 -0.2833  0.5575  0.218
  54  or  956  0.2567  0.5675 -0.3108  0.2512
  55  or  2521 0.07498-0.08733  0.1623  0.2696
  56  or  3600.008782 -0.4343  0.4431  0.2718
  57  or  2635 -0.1431  0.3721 -0.5152  0.1756
  58  or  2990 -0.1594 -0.3548  0.1954  0.2182
  59  or  1778 -0.2436  0.6652 -0.9088  0.1684
  60  or  3013 -0.3261 -0.2129 -0.1132  0.2016
  61  or  1318 -0.7963-0.01821 -0.7781  0.2565


fordchrist675 16-03-2010 14:17

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Quote:

Originally Posted by Bongle (Post 938076)
Mine failed from USFIRST but managed to use TBA as a backup.
Code:

Pos  Reg  Team    OPR    SAA      PM    RPI
  1  or  997    4.44  2.997  1.443  0.2869
  2  or  753  3.342  2.472  0.8707  0.2321
  3  or  3165  3.214  2.257  0.9566  0.2528
  4  or  1540  2.621  1.439  1.182  0.2585
  5  or  3192  2.537  2.571-0.03447  0.1906
  6  or  2910  2.168  0.6331  1.535  0.2883
  7  or  2865  2.045  2.172 -0.1265  0.1994
  8  or  948  1.979  2.207 -0.2276  0.2096
  9  or  2471  1.975    1.27  0.7048  0.2358
  10  or  1432  1.922  1.494  0.4277  0.213
  11  or  368  1.915  2.293 -0.3781  0.2302
  12  or    8  1.825  1.703  0.1226  0.271
  13  or  3145  1.814  1.986 -0.1718  0.2204
  14  or  957  1.791  2.599 -0.8078  0.1842
  15  or  847  1.709  0.6439  1.066  0.2785
  16  or  2811  1.701  0.1783  1.522  0.3023
  17  or  3223  1.692  1.6860.005676  0.2277
  18  or  2046  1.685  1.653 0.03189  0.1667
  19  or  1983  1.631  0.7979  0.8326  0.2701
  20  or  2557  1.466  0.9125  0.5539  0.3004
  21  or  488  1.444  1.262  0.1814  0.2307
  22  or  2733  1.373    1.67 -0.2969  0.2069
  23  or  2130    1.37  1.268  0.1024  0.2588
  24  or  3024  1.366  2.677  -1.312  0.1901
  25  or  1595    1.33  2.216 -0.8864  0.2402
  26  or  949  1.251  2.401  -1.15  0.2016
  27  or  1823  1.191  1.527  -0.336  0.2294
  28  or  2542  1.172  1.343 -0.1711  0.2331
  29  or  1700  1.089  1.579 -0.4898  0.2202
  30  or  3131  1.004  1.263  -0.259  0.1614
  31  or  3210  0.938  0.4483  0.4896  0.2581
  32  or  2374  0.914  0.3753  0.5387  0.255
  33  or  2951  0.8851  2.232  -1.347  0.1808
  34  or  1515  0.8275  0.6605  0.167  0.177
  35  or  2147  0.7307    1.34 -0.6097  0.2248
  36  or  2002  0.6813  0.3311  0.3502  0.2314
  37  or  1510  0.6691  0.7407-0.07164  0.2381
  38  or  2605  0.6604  1.997  -1.337  0.2139
  39  or  1425  0.5979  1.283  -0.685  0.2014
  40  or  2517  0.5943  0.2731  0.3212  0.2188
  41  or  3070  0.5522  0.4322    0.12  0.2002
  42  or  2898  0.5402  0.4317  0.1084  0.2305
  43  or  2522  0.5236  0.7541 -0.2305  0.199
  44  or  2192  0.5206  0.8281 -0.3075  0.2332
  45  or  955  0.5094  0.50780.001674  0.2162
  46  or  3311  0.4487 -0.1401  0.5888  0.1892
  47  or  2922  0.4001    1.29 -0.8898  0.221
  48  or  2411  0.3854  0.5043  -0.119  0.2797
  49  or  3188  0.3425  0.7973 -0.4548  0.2466
  50  or  2915  0.3389 -0.1069  0.4458  0.2668
  51  or  2550  0.2831  0.6419 -0.3587  0.2233
  52  or  3213  0.2749  0.5467 -0.2718  0.2069
  53  or  1571  0.2742 -0.2833  0.5575  0.218
  54  or  956  0.2567  0.5675 -0.3108  0.2512
  55  or  2521 0.07498-0.08733  0.1623  0.2696
  56  or  3600.008782 -0.4343  0.4431  0.2718
  57  or  2635 -0.1431  0.3721 -0.5152  0.1756
  58  or  2990 -0.1594 -0.3548  0.1954  0.2182
  59  or  1778 -0.2436  0.6652 -0.9088  0.1684
  60  or  3013 -0.3261 -0.2129 -0.1132  0.2016
  61  or  1318 -0.7963-0.01821 -0.7781  0.2565


Thank you oh so very much :D

The Lucas 16-03-2010 14:19

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Quote:

Originally Posted by Bongle (Post 938047)
It should just require the MSVC runtimes (I think, I'm not good at this redistributing thing), which I imagine are included in the .NET distributable. Probably won't help with your linux needs though, since it uses the Win32 WinINet functions rather than an OSS standard.

I called it OPRNet because it was (inter)net-capable, rather than having the user manually create text files to parse.

OK, that explains the name. Installing .NET is probably a simple solution for people having trouble running OPRNet (particularly on old XP laptops).

Also, I dont want/need a Linux version, the Windows version is great. Thanks Bongle! I have a dual boot machine and the Windows boot is what I use at competitions anyway to run Windriver, Excel and save my battery. Windows is the right choice since all of the KOP software is Windows.

cziggy343 16-03-2010 14:37

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Quote:

Originally Posted by billbo911 (Post 937845)
As requested.

thank you! :D

Jacob Plicque 16-03-2010 16:39

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Did the OPR miss the mark for the Florida Regional? The alliance of 1251 + 1612 + 86 won the whole shebang with a combined OPR of 5.4 versus alliances with group OPRs of 9.9, 9.2, 7.1, 6.4, 5.6, 5.2, and 5.10. If OPR were a good predictor wouldn't the team of 1592 + 179 + 3164 be a virtual lock with a group OPR of 9.9? In the past OPR usually was a good indicator of the strongest alliance in the elimination rounds. In the Quarterfinals, the top three teams by group OPR were knocked off.

billbo911 16-03-2010 17:10

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Quote:

Originally Posted by Jacob Plicque (Post 938187)
Did the OPR miss the mark for the Florida Regional? The alliance of 1251 + 1612 + 86 won the whole shebang with a combined OPR of 5.4 versus alliances with group OPRs of 9.9, 9.2, 7.1, 6.4, 5.6, 5.2, and 5.10. If OPR were a good predictor wouldn't the team of 1592 + 179 + 3164 be a virtual lock with a group OPR of 9.9? In the past OPR usually was a good indicator of the strongest alliance in the elimination rounds. In the Quarterfinals, the top three teams by group OPR were knocked off.

Great observation.
You need to understand, OPR is just one tool to help you decide. I'll bet in qualifications, 1251, 1612 and 86 were not able to compete along side teams that would compliment their game strategies or abilities. Look at it as a perfect storm. When you bring three teams together that fully complimented each others abilities and were able to play as a single unit, their performance as an alliance would be way better than they were able to show individually during qualifications.
This is why raw numbers is not always your best predictor. Observation and paying attention to all input is a scout's best approach. Leaning on one detail, like OPR, can be mis-leading.

Bongle 16-03-2010 17:22

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Another reason OPR might be a poor elimination predictor is because the rules essentially change in eliminations.

In qualifying, it isn't really in anyone's interest for teams to play heavy-handed defense. In eliminations, defense is a key factor. So teams with a pneumatic-tire 8-motor, rocket-powered 8WD suddenly are much more useful, while teams with highly mobile feather-light (and light on grip, like mechanum/omni) robots suddenly find it much harder to score.

A perfect example is 469: they have a low OPR (well... compared to their reputation) because they are only at maximum effectiveness when they're playing with highly effective robots that can get their ball loop going. In qualifying, that might not happen often. An ball-supplier bot is limited in offensive power by its home-zone teammate that is trying to get balls in the net. 469 is a defense-proof near-perfect ball supplier. When defense ramps up and solid ball-deliverers become available, suddenly 469 is an unstoppable force.

Jacob Plicque 17-03-2010 21:58

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
At the Oregon, New Jersey, and Pittsburgh regionals, the alliances with the best OPR were the winners. Over the years, it has been consistant that the OPR team score was a 90%ile indicator of sucess. Bongle makes a great point about the defense as the 1251+1612+86 alliance was the number 2 rated defense in the eliminations. I am curious about how many of the regional winning teams were predicted by OPR for weeks 1 & 2 in 2010. At a glance, Florida seems the exception.:confused:

Mr. Lim 17-03-2010 22:12

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Also, OPR cannot take into account teams that are playing the "seeding points game" by scoring goals for their opponents. There are several very high performing teams I know of that have abysmal OPRs for this reason - because 2 or 3 times a match they were scoring for the other alliance. That's a huge hit to your OPR, SAA and ultimately your PM.

Nawaid Ladak 17-03-2010 22:30

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
I think you need to take a look at all four stats to determine who the best robot is on the field.

High OPR + Unusually High SAA + mediocre PM + Strong RPI = a really good robot
High OPR + Low SAA + strong PM + Strong RPI = Good Robot

thats how i look at the stats.

Jacob Plicque 18-03-2010 00:17

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
I have not used RPI in the past since it has such a narrow range of data like 0.16 to 0.40. Obviously higher is better as an indicator of strength of schedule and wins. However its relationship to OPR, DPR, & PM is not easy to compare since these valus are often 20 times larger

Bongle 22-03-2010 12:54

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
I noticed this weekend that this is the year of the prediction feature. Running it for KC, the self-check indicates it would have been 70% correct after only 39 matches, and consistently 80% correct after 48 matches. For Lunacy it does much worse. This seems to indicate that this game is much more predictable, and that good robots in one match will often do well in subsequent matches. Note that this is only for predicting the winner. So although it is better at predicting the winner of a match than last year, that's a less useful thing to do than it was last year.

Code:

After 10 matches, OPR would not have been computable
After 11 matches, OPR would not have been computable
After 12 matches, OPR would not have been computable
After 13 matches, OPR would not have been computable
After 14 matches, OPR would not have been computable
After 15 matches, OPR would not have been computable
After 16 matches, OPR would not have been computable
After 17 matches, OPR would not have been computable
After 18 matches, OPR would not have been computable
After 19 matches, OPR would not have been computable
With 20 matches of data, match prediction would have been 50% of the time
With 21 matches of data, match prediction would have been 56% of the time
With 22 matches of data, match prediction would have been 48% of the time
With 23 matches of data, match prediction would have been 63% of the time
With 24 matches of data, match prediction would have been 62% of the time
With 25 matches of data, match prediction would have been 55% of the time
With 26 matches of data, match prediction would have been 58% of the time
With 27 matches of data, match prediction would have been 50% of the time
With 28 matches of data, match prediction would have been 67% of the time
With 29 matches of data, match prediction would have been 50% of the time
With 30 matches of data, match prediction would have been 50% of the time
With 31 matches of data, match prediction would have been 52% of the time
With 32 matches of data, match prediction would have been 61% of the time
With 33 matches of data, match prediction would have been 69% of the time
With 34 matches of data, match prediction would have been 64% of the time
With 35 matches of data, match prediction would have been 60% of the time
With 36 matches of data, match prediction would have been 66% of the time
With 37 matches of data, match prediction would have been 66% of the time
With 38 matches of data, match prediction would have been 65% of the time
With 39 matches of data, match prediction would have been 71% of the time
With 40 matches of data, match prediction would have been 71% of the time
With 41 matches of data, match prediction would have been 77% of the time
With 42 matches of data, match prediction would have been 80% of the time
With 43 matches of data, match prediction would have been 75% of the time
With 44 matches of data, match prediction would have been 76% of the time
With 45 matches of data, match prediction would have been 74% of the time
With 46 matches of data, match prediction would have been 73% of the time
With 47 matches of data, match prediction would have been 75% of the time
With 48 matches of data, match prediction would have been 82% of the time
With 49 matches of data, match prediction would have been 80% of the time
With 50 matches of data, match prediction would have been 83% of the time
With 51 matches of data, match prediction would have been 87% of the time
With 52 matches of data, match prediction would have been 87% of the time
With 53 matches of data, match prediction would have been 78% of the time
With 54 matches of data, match prediction would have been 80% of the time
With 55 matches of data, match prediction would have been 77% of the time
With 56 matches of data, match prediction would have been 74% of the time
With 57 matches of data, match prediction would have been 73% of the time
With 58 matches of data, match prediction would have been 75% of the time
With 59 matches of data, match prediction would have been 75% of the time
With 60 matches of data, match prediction would have been 84% of the time
With 61 matches of data, match prediction would have been 81% of the time
With 62 matches of data, match prediction would have been 83% of the time
With 63 matches of data, match prediction would have been 80% of the time
With 64 matches of data, match prediction would have been 88% of the time
With 65 matches of data, match prediction would have been 88% of the time
With 66 matches of data, match prediction would have been 84% of the time
With 67 matches of data, match prediction would have been 81% of the time
With 68 matches of data, match prediction would have been 87% of the time
With 69 matches of data, match prediction would have been 86% of the time
With 70 matches of data, match prediction would have been 86% of the time
With 71 matches of data, match prediction would have been 85% of the time
With 72 matches of data, match prediction would have been 85% of the time
With 73 matches of data, match prediction would have been 84% of the time
With 74 matches of data, match prediction would have been 84% of the time
With 75 matches of data, match prediction would have been 83% of the time
With 76 matches of data, match prediction would have been 82% of the time
With 77 matches of data, match prediction would have been 81% of the time
With 78 matches of data, match prediction would have been 80% of the time
With 79 matches of data, match prediction would have been 80% of the time
With 80 matches of data, match prediction would have been 78% of the time
With 81 matches of data, match prediction would have been 77% of the time
With 82 matches of data, match prediction would have been 82% of the time
With 83 matches of data, match prediction would have been 81% of the time
With 84 matches of data, match prediction would have been 80% of the time
With 85 matches of data, match prediction would have been 78% of the time
With 86 matches of data, match prediction would have been 76% of the time
With 87 matches of data, match prediction would have been 75% of the time
With 88 matches of data, match prediction would have been 81% of the time
With 89 matches of data, match prediction would have been 80% of the time
With 90 matches of data, match prediction would have been 77% of the time
With 91 matches of data, match prediction would have been 87% of the time
With 92 matches of data, match prediction would have been 85% of the time
With 93 matches of data, match prediction would have been 83% of the time
With 94 matches of data, match prediction would have been 80% of the time
With 95 matches of data, match prediction would have been 100% of the time
With 96 matches of data, match prediction would have been 100% of the time
With 97 matches of data, match prediction would have been 100% of the time
With 98 matches of data, match prediction would have been 100% of the time


Bongle 23-03-2010 20:12

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
2 Attachment(s)
Man this code is awful.

Anyway, v12 (based on v7) is now ready. The prediction feature now has awareness of the new seeding system, though since it can't know penalties, the predicted seeding scores are too high. Even if I gave it an entire regional of match scores with no prediction, the rankings it would give out would still be incorrect because teams would not be getting as many points as they should.

Clinton Bolinger 24-03-2010 01:46

Re: Easy to use Offensive Power Rankings (OPR) program for mid-regional scouting
 
Looks like you are missing the Ann Arbor District Event (WC is the abbreviation that FIRST uses) Last year it was the Lansing District.

-Clinton-


All times are GMT -5. The time now is 00:38.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi