paper: 2006 frc195 Statistical Team Analyzer

[Tom Bottiglieri]

Thread created automatically to discuss a document in CD-Media.

2006 frc195 Statistical Team Analyzer (http://www.chiefdelphi.com/media/papers/1830?) by

[Greg Marra]

It's too bad that FIRST hasn't released the data for the qualification matches at UTC. Other than missing part of the dataset, you've put together a pretty sweet combined scouting sheet here.

Let's see how it holds up at the event! :D

[Lil' Lavery]

Contains the same errors that 1114's whitepaper has (mainly the St. Louis elimination rounds, for instance 71 won the regional, while both papers state they didn't make the elimination rounds, but also the Peachtree and Souther Cal finals were inverted in both as well), but overall a terrific job! Another fine tool for pre-championship scouting!

[Karthik]

Tom & Team 195,

I love what you've done here. You've created a very useful and "smart" metric. It's always good to see teams using the power of statistics to further their analysis. Great work.

[Bill_Hancoc]

I olny found one problem-Some of the teams that are on the TEAM DATA page are not listed in the Raw Data page and therefor do not show up when entered on the Inerface page. I dont know if this is due to them not going to ATL but just pionting that out. I dont know how many teams this applies to but i know that team 7 is one of them.

[Lil' Lavery]

I beleive that is because they arn't going to Atltanta. 007 is not going this year, for example. Other teams that are on the team data page but not raw data include; 116, 117, 414, and 1731, none fo which will be competing at the Championship event this year.

[Tom Bottiglieri]

I olny found one problem-Some of the teams that are on the TEAM DATA page are not listed in the Raw Data page and therefor do not show up when entered on the Inerface page. I dont know if this is due to them not going to ATL but just pionting that out. I dont know how many teams this applies to but i know that team 7 is one of them.
The 'Team Data' page includes all teams in FIRST. The 'Raw Data' page is taken from Karthik's database, and only includes teams going to ATL.

[Bill_Hancoc]

Thanks for clearing that up, very nice job

[Tom Bottiglieri]

I don't know if this is fixable, but for QF Match 35 at BAE, the score was recorded wrong. According to this video of that match, the final score was 53 to 53. According to what FIRST posted on their website, the final score was 5 to 5. Seeing as we only played 9 qualifying matches, that raises our score by 5.3 points.

Ill update that when I throw in the UTC data tomorrow. I have another workbook with all of the regional match data that generates the 'Team Data' sheet. I would post that book as well but it is huge (over 5mb) and all of the macros are extremely.. well.. for lack of better words, terrible. :cool:

[Spindash54]

Hey, this looks amazing. One little thing though.

Our team is only mentioned in the Team Data center, but not the Raw, Curie Rank, or Interface.

I look foward to using this data down at the championship.

[Tom Bottiglieri]

Hey, this looks amazing. One little thing though.

Our team is only mentioned in the Team Data center, but not the Raw, Curie Rank, or Interface.

I look foward to using this data down at the championship.
Ahh.

I used the original division lists which did not include the teams who were pending. (103 for example)

I'll make sure to add these teams to the database tomorrow. If anyone else sees any errors, please post them here.

[Tom Bottiglieri]

I took about 3 hours today manually adding in match data from UTC based on our scouting records.

Revision 2 is available on the download page for this paper. Just click the link in the first post.


Problems solved:

The UTC data has been added to the overall database, and UTC only teams have been incorporated into their division listings
Division team listings have been updated to include teams who were "pending"


Adding the UTC info has had a small influence on the rankings. Some teams moved around, and some were added in high on the ranks (top 15 - 20,195)

Enjoy!

[Joel J]

Hi.. Can I get that UTC data!?

[Ben Piecuch]

Geez, thanks Tom! (please note some sarcasm...) We, team 228, went from one of the mysterious unrated Curie team to a -4.7 average score. Ouch!

Just to let all the Curie scouts know, we did not have a working ball delivery system for UTC. And therefore, were only able to shot about 2 balls a match before the whole thing jammed. We've designed and tested a fix for the problem, and will hopefully have it installed and working by mid-Thursday. Please stop by our pits to say hi, and to take a look at the new system. Best of luck to all the teams!

BEN

[Tom Bottiglieri]

Geez, thanks Tom! (please note some sarcasm...) We, team 228, went from one of the mysterious unrated Curie team to a -4.7 average score. Ouch!

Just to let all the Curie scouts know, we did not have a working ball delivery system for UTC. And therefore, were only able to shot about 2 balls a match before the whole thing jammed. We've designed and tested a fix for the problem, and will hopefully have it installed and working by mid-Thursday. Please stop by our pits to say hi, and to take a look at the new system. Best of luck to all the teams!

BEN
Sorry Ben :cool:

I think I know the source of your problem. Since our scouting team packed up after Friday's matches, the UTC data doesn't include your last 2-3 matches. These were probably the matches you did well in, and how you ended up with your #10 rank.

And yes, 228 has a great bot this year. They had some problems with their shooter early on, but were able to pull off some good plays by the end of UTC. I expect nothing less than perfection with their new fix.

[sw293]

Interesting & good job. I have a couple of small questions:

Why should the average scale score at each regional be zero? It seems that this would penalize teams that attended stronger regionals when comparing scores across regionals.

Also, did you try scaling the scores once over the entire season, i.e. treating the entire FIRST season as one regional and scaling the scores based on that? (I realize there are plenty of reasons for not doing this).

[AndyB]

you are my hero. amazing job!

[Tom Bottiglieri]

Revision 3:

Lil' Lavery fixed the event history section for St. Louis, SoCal, and Peachtree.

[DanDon]

Hey Tom,

I don't mean to ask for more work from you, you did an awesome job already, but is there any way for you to make a sheet with the teams ranked in divisions based on highest scaled score?

If not, its no biggy,

[Compnerd]

Are you going to do a 2007 analyzer?

[Tom Bottiglieri]

Are you going to do a 2007 analyzer?
Probably. It depends on the match data that FIRST archives. Without a good sample size, the data is worthless.