Talk:DEWBOT VI Scouting
Contents
IRI Scouting Team
Sign up as desired
Preliminary Research
Collects, compiles, and analyzes data on attending teams' previous performance from scouting databases, match archives, and forum information. (Occurs before the competition)
Pit Interviews
Talks to attending teams in their pits (at the competition) to gain insight into their team and robot characteristics and performance. Great for design insights and ideas.
Match Observation
Watches competition matches and records relevant data. (Requires a laptop, please bring a flash drive. Inform Siri if you cannot bring a laptop.)
Unassigned
Assigned as needed (and as available) throughout competition.
Scouting Continuous Improvement Discussion
Welcome to the FRC Scouting CID! Please feel free to leave any comments or suggestions on information you think should be scouted. This includes pre-event preliminary research and at-event match observation & pit interviews. Also feel free to comment on layout/user interface possibilities (printed and digital). Please sign your entries (with--~~~~).
| Deadlines for Suggestions: IRI Preliminary Research - 26 June, 2010 |
Current Quantitative Scouting Metrics
Preliminary Research
- Team number
- Team nickname
- Team website (hyperlink, digital versions only)
- Rookie year
- Number and names of previous official competitions (divisions separate from Einstein) this season
- Number of team attending each of the above
- Performance and engineering awards this season, including multiplicities
- Draft order for each elimination placing
- Team 1114 Stats for last regional this season
- Calculated Contribution (World Percentile)
- Average Offensive Score (World Percentile)
- Seeding Score (World Percentile)
- Average Defensive Score (World Percentile)
- Win-loss-tie, absolute & percentile (100% high) rank for each official competition this season
- Recent awards
- names listed for 2 years before current season,
- or for perennial winners, numerical award:years ratio back to 2001 or rookie year (whichever is later)
- Average awards/year for latter
- Notes
The current quick color code system is: "Scary" (e.g. division placers, regional winners/multiple finalists, etc); "Important" (e.g. regional finalists and semifinalists, high draft quarter-finalists); "Notable" (e.g. regional quarter-finalists, multiple engineering award winners); and "Mentionable" (e.g. engineer award winners, high 1114 and/or W-L-T stats). Examples are very general and anything from some CD posts to incongruous 1114 stats can affect a team's code. The colors shouldn't replace carefully studying the rest of the scouting material.
Planned Scouting Metrics (IRI)
TBA 27 June for Preliminary Research and 5 July for Match Observation & Pit Interviews
Suggested Changes
[Siri:]
Preliminary Research
Description: Collecting, compiling, and analyzing data on attending teams' previous performance from scouting databases, match archives, and forum information.
Purpose: Initially sort
- Add average score & winning margin from last regional.
- Rationale: Separates the lucky-to-be-picked quarter-finalists from the unlucky-to-lose quarter-finalists, arguably even better than draft order.
- Drop all but last regional & championship performances, including average.
- Rationale: The next competition is in July, I don't much care how teams did at their first regional in March. Does anyone use these stats anymore?
- Add important data from major off-season competitions.
- Rationale: This information is more current. Problems: Inconsistency/misinterpretation of data, especially for competitions we don't know (very likely with IRI teams). Also, lack of data for some teams and huge time commitment. This will be used only to provide more information on seemingly contradictory records.
- Add notes on hanging ability. (and record?)
- Rationale: This is proves important, not just for knowing how well an alliance member can hang, but for guessing what opponents will do in the finale.
- Add scoring and defending ability from each zone (in auton & teleop for former) from prior competitions.
- Rationale: exceedingly helpful, but subjective and really hard to find from 72 teams across the world.
- Drop recent (non-2010) award results, except for those who present as good for networking/advice.
- Rationale: This was the original purpose of the category, which has sort of spiraled into less-usefulness.
- Suggested Awards to List: Chairman's, Autodesk Design, Engineering Inspiration, Entrepreneurship, Website
- Drop statistics on the number of attending teams at teams' previous competitions.
- Rationale: This is a lot of work for not much information, especially at competitions like IRI (lots of teams, large geographical area). Seeding percentile will come from team 1114's stats (calculated the same way it has been).
- Add CCWM World Rank percentile (calculated from 2834's stats).
- Rationale: helpful, especially in mitigating the difficulty with FiM-to-traditional comparisons.
- Add percent of qualification matches won.
- Rationale: This is more compact, efficient and telling than W-L-T record.
- Use qualification instead of total W-L-T record.
- Rationale: This is a more accurate comparison that the overall record (since elimination matches are more difficult).
- Use overall championship rank percentage instead of divisional.
- Rationale: This is a more accurate comparison across divisions, however it can give false negatives. (Instead of 99% being the "absolutely horrifying" cutoff, it's roughly 92%.)
- Bold average defensive values over 89.5 as well as the Michigan State Championship (in the "last regional" column).
- Rationale: The former brings attention to outstanding and unexpected D values (which average at 60 for IRI, versus 80 for most other metrics). The latter brings attention to the difference between traditional regionals and the Michigan State Championship (which teams must qualify for).
- Differentiate between Michigan district awards, Michigan State Championship awards, and Michigan Championship awards.
- Rationale: Michigan vs. traditional comparisons are difficult--information is power.
- Indicate 1114 stats that are 100 exactly.
- Rationale: This is ridiculous and should be acknowledged.
Requirements
- Good video archives (especially difficult for off-seasons)
- 1114 & 2834 Database integration (preferably automated?)
- Time and patience
Match Observation
Description: watching competition matches and recording relevant data.
- Ball manipulation ability
- Rationale: Same as above, but way easier to do on-site for a given competition. We should do this! (and keep the records!)
- Number of balls scored from each zone in teleop (Priority 1) & auton (P2)
- Number of balls cleared over n bumps in teleop (P3) & auton (P4)
- Defense ability & tactics
- Rationale: Same as above. We should do this too!
- Defensive strategy (clear balls, block bots, push, stuff in goal, whatever) (P5)
- Defensive success (did it work? number of balls scored on them, etc) (P5)
- Autonomous success/tactics (do they move? what pattern(s)? if they don't score, how close are they?) (P6)
- Ease/speed on bump (P7) & tunnel (P9)
- Rationale: Helpful when they're trying to come to your rescue. More helpful when they're trying to shut you down.
- Hanging success (P8) & approximate time (P10)
- Rationale: Same as hanging stats.
- Starting Zone (P11)
- Rationale: Helpful to know for opposing teams when selected allied starting zones.
- Major zone changes/ending zone (P12)
- Rationale: Fewer surprises to prepared alliance strategy.
Requirements
- n trained & observant scouts per match. n=6 for P1-P12, 3+ for P1-P5
- Easy-to-use (preferably without looking) UI
- Easy-to-operate database/readout
Pit Interviews
Description: talking to attending teams in their pits (at the competition) to gain insight into their team and robot characteristics and performance.
- Ball manipulation methods
- Rationale: There's no such thing as too much benchmarking!
- Kicking methods & drivetrains
- Rationale: ...Still can't have too much benchmarking.
- Tangential Award Success (Chairman's, Autodesk, Website, EI, Entrepreneurship)
- Rationale: Always room for improvement (and making friends)
- Driver Experience
- Rationale: Big factor at many off-seasons, it's definitely good to know when and if the experienced drivers step up to the controls.
- Current changes/improvements
- Rationale: This is subjective (and sometimes embellished), but things do change fast in FIRST. Helps to have an idea of where to look hardest.
Requirements
- Friendly & prepared interviewers
- Easy-to-use & easy-to-integrate (into database) interview sheets
Other Goals
- Explain why you should pick us
- Make friends & contacts
- Learn/observe other good practices
- Liaise with judges (important for anyone, especially students, in the pit)
--Siri 12:06, 9 June 2010 (UTC)
Available Scouting Metrics (Not Currently in Use)
Preliminary Research
- (FIRST) Location
- (FIRST) Photo and colors (some)
- (FIRST) Full Team Name
- (FIRST) Qualification Win-Loss-Tie Record by Regional
- (FIRST) Total Qualification W-L-T Record
- (FIRST) Seeding Score by Regional
- (FIRST) Coopertition Bonus by Regional
- (FIRST) Hanging Points by Regional (note: names can be deceiving. 1640's hanging score was 4.00 at Drexel--pretty good for a robot without a hanging apparatus that never used a ramp in competition.)
- (FIRST) Qualification Matches Played by Regional
- (FIRST) Individual Match Results by Regional
- (FIRST) All Awards by Regional and by Team
- (2834) Average Winning Margin Per Match by Field
- (2834) Calculated Contribution to Winning Margin by Regional
- (2834) Overall CCWM
- (2834) CCWM Regional Rank
- (2834) CCWM World Rank
- (2834) Average Score per Match
- (2834) Offensive Power Rating by Regional
- (2834) OPR Regional Rank
- (2834) OPR World Rank
- (1114) Team's Average Offensive Score
- (1114) Regional's AOS
- (1114) AOS Regional Rank & Percentile
- (1114) AOS Champions' Rank & Percentile
- (1114) AOS World Rank & Percentile
- (1114) All of above 1114 with Calculated Contribution, Seeding Score & Average Defensive Score
- (Theoretical) It's theoretically possible to derive a team's average penalties (from overall alliance by match), hanging bonuses & goals from the Twitter feed, but I don't know how to do this (even if we decided we want the information). --Siri 18:02, 7 June 2010 (UTC)
