Final Results
Interactive and Visual Statistics for Volleyball
Advisor: Eric Alexander
Background
While data-driven analysis has always had a place in sports, it has experienced an upturn in recent years. Coaches and managers are using increasingly advanced analysis tools to guide their decision making, while films like Moneyball and sites like FiveThirtyEight are making statistics cool in the eyes of the fans. However, while professional leagues have the staff and the funding to track the actions and location of each player at all times and calculate advanced metrics, teams at other levels are limited in the data they can collect and the analysis they can perform with it.
At the D-III collegiate level, many stats are taken about the performance of both opposing teams and one’s own team. Most of this data is taken by hand--some mid-game, some while watching film after the fact--and must then be later collected, aggregated, and analyzed (see an example below). As such, it is very difficult for coaches to make decisions based on this data in the moment--to act upon it when it is most relevant. Interactive tools for data collection, analysis, and presentation could change this.
This project will focus on data-driven analysis of volleyball. Volleyball presents an interesting challenge, for while the actions taken by the players are diverse, there are enough constraints on those actions (where players can start, the discretization of games into points, etc.) that it may be possible to move beyond computing box-score style statistics to create models of player and/or team behavior.
The project
For this project, you will develop a tablet app that can be used by the coaches and staff of the Carleton volleyball program. This app will be responsible for affording both the collection of in-game statistics and the presentation of data that can inform decision making in real time.
Important pieces of this project will include:
- Problem analysis: What are the questions that coaches want to ask of the data? This will involve working in collaboration with Coach Heidi Jaynes and her staff and players to develop a set of needs and requirements.
- Data analysis: What models will help answer these questions? How can these models be updated to reflect new data in real time? This will require research into the different algorithms that can be employed to compute aggregate statistics, cluster types of player behavior, etc.
- Data collection: How can data be collected both accurately and efficiently while play is happening? This will involve iterative design and development of an interactive tablet interface.
- Data presentation: How can models of the data and recommendations be presented in a way that will support in-game decision making? This will likely warrant the design and implementation of visualizations to convey relevant data or models.
Recommended experience
This project will involve a combination of working with data and working with humans. Previous experience in extracting insight from numerical data (e.g., CS 32X, Data Visualization, statistics) will be valuable, as will experience in human-centric design (e.g., Software Design, Human-Computer Interaction, Mobile Application Development). All of this experience need not reside in any single person! Most important is an enthusiasm for bridging this gap, as well as hopefully an interest in the problem domain.