Re: paper: New Scouting Database from Team 2834
 

For those who didn't want to spend the time to download, unzip and open the file, here is the top ten true world OPR ranking at the end of week 2. It takes into account all qualifying round matches of every regional and district. There are enough teams that have played two events already so there are enough interactions that make the true world OPR different from the OPR calculated from just one event.

Re: paper: New Scouting Database from Team 2834
 

Week 3 results are included.

In order to maximize interactions between teams in the calculations of World Ranking, the elimination rounds were included. I think this approach is valid since some teams attend more events than others anyway. The more matches we take into account, the more accurate the results will be.

For individual districts and regionals, only qualifying matches were used to calculate OPR and CCWM in order not to skew the results. This is consistent with the belief of the CD community.

The following is the top ten World Ranking.

Re: paper: New Scouting Database from Team 2834
 

Ed,
There is something wrong with the world rank sheet. It shows Team 86 with 16 wins and 6 losses. We only have 9W-6L with the playoff games. Our only competition was the Florida Regional.
Jacob
:confused:

Re: paper: New Scouting Database from Team 2834
 

Jacob,

Thanks for catching the mistake. All the data of the World Ranking is correct except for the column Wins (column C). The program overwrote that column with the number of matches played (column F) by mistake which is why the two columns had the same value. I have posted version 3 to correct the mistake.

Re: paper: Scouting Database from Team 2834
 

Here is the corrected Top 10. Again, only the column Wins posted before was wrong.

Re: paper: New Scouting Database from Team 2834
 

The scouting database has been updated to include Week 4 results.

The top 10 true world ranking is as follows.

1114 continues to increase their OPR and CCWM and is now ranked #1 in both. Their MP is an amazing 150 from a match this weekend.

Regional Rank: New Scouting Database from Team 2834
 

Ed,
Your data is also useful in determining the strongest regional. Using the average OPR of the top 24 robots a strength rating is created as follows:

1 Midwest, 28.02
2 Oregon, 26.98
3 WashingtonDC, 26.12
4 LosAngeles, 25.09, 17.41, Average
5 AnnArbor, 23.28, 4.76, Std Dev
6 KansasCity, 22.58, 31.71, +3 Std Dev
7 WestMichigan, 21.40, 3.12, -3 Std Dev
8 FingerLakes, 20.5
9 NewJersey, 20.48
10 Wisconsin, 20.15
11 SanDiego, 20.07
12 BAE, 19.9
13 Chesapeake 19.05
14 Niles 19.00
15 Pittsburgh 18.16
16 Sacramento 17.92
17 Waterloo 17.78
18 Alamo 17.8
19 Palmetto 17.73
20 LoneStar 17.19
21 Detroit 16.68
22 Boilermaker 16.61
23 Florida 16.25
24 Oklahoma 15.8
25 NewYorkCity 15.37
26 WPI 15.21
27 LongIsland 15.18
28 Bayou 14.9
29 St.Louis 13.97
30 Kettering 13.6
31 SeattleCascade 12.52
32 LakeSuperior 12.4
33 Peachtree 12.1
34 Waterford 11.86
35 SeattleOlympic 10.8
36 Israel 10.4
37 TraverseCity 9.97
38 Arizona 9.06
:D

Re: paper: New Scouting Database from Team 2834
 

Jacob,

I have seen studies done with several years of data that shows the average OPR increasing as time goes on so I would be careful in saying one regional/district is stronger than another if they are not in the same week.

Take our team as an example. Our week 1 event OPR was 5.9 because the robot was not working properly. Our week 2 event OPR was 32.5 because the robot was working and we also got the minibot to work half way through the event.

For teams that do not have a practice robot to practice after ship date, their OPR usually improves as drivers get more experience. There can be many other reasons teams improve their OPR from one event to the next. Most teams plateau out after their 3rd event. Of course some teams' OPR goes down as their robot wore out and start to break down.

One difference between powerhouse teams and teams like ours is they start their season with very high OPR probably because they have less things to "debug". For us we still have a lot of things to learn. We have a lot of "first" this year. First time using banebot motor, first time using gyro, first time using potentiometer, first time using Mecanum wheels and programming it, first time programming a field oriented drive, first time designing and building a 4 bar link arm, first time trying a magnetic encoder, first time trying a compass sensor. We are very lucky to have Team 33 helping us but when you are doing so many things for the first time, bad things are going to happen.

Regional Ranking: New Scouting Database from Team 2834
 

Ed,
I agree that my method does not reflect improvement from Regional to Regional. However, 69.5% of the teams only attend one regional, 28.9% of the teams attend two regionals, and 1.6% attend three or more regionals. The quick and dirty list does show dominate regionals that are ~3 times stronger than others. The use of the world rank data would probably produce a fairly accurate strength estimate. While the correlation is unproven, I'll bet the the winners of stronger regionals are dominating in the FIRST Championships.
:cool:

Re: paper: New Scouting Database from Team 2834
 

can you email me at tomasicenogle@gmail.com with the definitions of every part? EX: RS OPR ect... can i just get a list of everything? Thank you.

Re: paper: New Scouting Database from Team 2834
 

QS and RS definition can be found in Section 5 of the manuals.

5.3.3 Qualification Score (QS)
Qualification points are awarded to each TEAM at the completion of each qualification match and are dependant on the final score:
 Each TEAM on the winning ALLIANCE will receive two (2) qualification points.
 Each TEAM on the losing ALLIANCE will receive zero (0) qualification points.
 In the event of a tied score, all six TEAMS will receive one (1) qualification point.
The total number of qualification points earned by a TEAM throughout their qualification matches will be their qualification score.

5.3.4 Ranking Score (RS)
Each TEAM on the winning ALLIANCE will receive a number of ranking points equal to the unpenalized score (the score without any assessed penalties) of the losing ALLIANCE.
Each TEAM on the losing ALLIANCE will receive a number of ranking points equal to their final score (with any assessed penalties).
In the case of a tie, all participating TEAMS will receive a number of ranking points equal to their ALLIANCE score (with any assessed penalties).
The total number of ranking points earned by a TEAM throughout their qualification matches, divided by the number of matches played (excluding any SURROGATE matches), then truncated to two decimal places, will be their ranking score.
Note: because your ranking score is derived directly from the match scores of the losing ALLIANCES in the matches you play, it is in your best interest to support your opponents and win by helping each ALLIANCE score as many points as possible.

OPR stands for Offensive Power Rating. CCWM stands for Calculated Contribution to Winning Margin. The explanation can be found in this white paper. http://www.chiefdelphi.com/media/papers/2174
Look for the file "Team 2834 Scouting Database Presentation 2010.pdf"

Re: paper: New Scouting Database from Team 2834
 

Is there any reason why I can't seem to populate the CT match list correctly? I get times and team numbers and scores for Qualifications but they seem random....

edit: It also seems like the Qualification data that it is pulling is going where the elimination data normally is

Re: paper: New Scouting Database from Team 2834
 

The links for future events are still pointing to last year's events. I cannot setup the new links ahead of time when the web pages did not exist until Thursday or Friday. This was possible before in Excel 2003, but this does not seem to be possible with Excel 2007 and 2010. This will be my summer project if I have time to automate this as much as possible so I can set them up quickly. Doing it manually is quite tedious when there are so many events.

People who knows Excel well enough are able to change the links themselves to point to the current data.

Re: paper: New Scouting Database from Team 2834
 

Based on requests, I have set up the links for Connecticut, Smoky Mountain and Troy for people to use. It is in version 5b. Good luck at the competition this weekend.

Keep in mind that the OPR numbers will appear very low until more matches are completed. But you can kind of scale it up based on number of matches played.

Re: paper: New Scouting Database from Team 2834
 

Here is Version 5c that takes care of the low predicted OPR issue but it will require mroe matches played before it will work. Otherwise the matrix is ill conditioned and you will get an error message.

Edit: It works for all 3 locations now.