Open realjoshuau opened 10 months ago
No assignees, but everyone should be working on this ideally.
Do you have an example of the types of data we collected in previous years?
maybe just add functionality to the app that makes it easy to just add new data collection type, similar to a google form? That way we don’t need to actually change any of the code: to add a new data point we can just add a new type of data from a drop-down menu, add some description whatever, and allow for weight changing in however the data points will be weighted depending on that year’s game.
*repost from teams
Do you have an example of the types of data we collected in previous years?
Only if we've made a previous scouting app. We usually collect ~3 types of data though.
Qualitative scouting (doesn't really depend on the game)
Quantitative scouting (depends on the game)
Pit scouting (50/50. we asked if they have a certain component X on their robot, i.e like a buddy climb)
https://github.com/Team3256/infinite-recharge-scouting-app/blob/main/my_wb_server/main.py https://github.com/Team3256/scouting-2023/blob/main/packages/db/prisma/schema.prisma https://github.com/Team3256/mywb-scouting-data-parser/blob/master/scouting_example.json
@memori034 can help you as well with this, since we used Google Forms during CalGames.
yeah in the interest of robot game qualitative match scouting:
judgement of performance during a match
ex: "played good defense against 3256 by preventing them from making efficient cycles through single substation" or "Driver was not very aware, having bumped into their alliance team members multiple times." quantitative match scouting:
numerical observations
ex: how many points scored in TeleOp period grid, or Autonomous charge station pit scouting:
mix of both qualitative and quantitative
various types of information that could prove useful in match strategy and alliance selection
ex: "robot weighs 125lbs with bumpers", "control system is a ti84", or "this team strongly prefers to score cubes"
@memori034
@realjoshuau do you think we can quantify qualitative data? or even have chatgpt rank/sort qualitative data? or did you plan on manually reading every qualitative comment as that sounds somewhat tedious and subjective.
we can do ratings on scales
@realjoshuau do you think we can quantify qualitative data? or even have chatgpt rank/sort qualitative data? or did you plan on manually reading every qualitative comment as that sounds somewhat tedious and subjective.
we probably can, qualitative data isn't just notes and long-form data (even though it is useful during picklist/match strat), but also relative ratings on scales (which is why we assign people experienced with scouting on qual scouting)
using AI to rank/sort qual data is interesting and I think we should 100% look into it. we need to train/make the model understand how frc works tho
@Saamstep lmao
@Saamstep lmao
@BK1031 💀😭
fire
Brainstorm ways that we could create a "scouting schema", i.e a way that we should specify what data could be collected, what transforms we could apply to the data, and how that data feeds into other functionality
Since we plan to reuse the app in the seasons coming, a "scouting schema" could be a way that we reuse the app by just re-specifying what data we want to collect.