Last year, I worked on a data science project based on Graham Allison’s Thucydides Trap idea. This year, I am going to do a different kind of project which looks at the FIRST® Robotics Competitions (FRC).
I have been with my high school’s robotics Team 341 Miss Daisy, since 9th grade. The 2022 FRC game is called Rapid React. It’s an exciting robotic game of shooting balls and climbing bars. Our team participated in three regional competitions before we went to the World Championship in Houston. Each FRC competition consists of two parts, the qualification round and the playoff round. During the qualification round, each team is randomly assigned with two other teams to constitute an alliance and plays against another randomly constituted alliance. Each alliance is scored based on various performance metrics at each game. At the end of the qualification round, each team is ranked by their cumulative or average performance scores. The top seeded teams get to choose their alliance partners to compete in the playoff round.
I’m interested in the strategy on how to pick alliance partners for the playoff round based on the performance during the qualification round. Traditionally, our Team 341 is known to do good scouting: we have team members watch every game during the qualification round; we collect additional performance data on each individual team. For this data science project, I also collect the official FRC competition data from the The Blue Alliance (https://www.thebluealliance.com/) website and our own team’s scouting data from the 2022 season. I hope to make some sense out of the data and find a way to automate our alliance selection process for future years.
For last year’s data science project, I learned different models for time-to-event data. For this year’s project, I plan to learn logistic regression. I did last year’s project in R. I think I can still use R to do the logistic regression. So far, I have read different data sets into R and manipulated them into the shape I want in R.
