The project proposals listed here are provided by external connections.

Please let Steven know if you are interested, to keep count, and then feel free to reach out to the sponsor directly.

Squash League Management

Contact Information

Contact: Jillian Anderson jillian_anderson@sfu.ca

Background

Jillian is currently (Jan 2024) the co-chair of Squash BC's doubles sub-committee. At a recent meeting, they were discussing the frustration the organization (and players) have with the Doubles Rating System used by Club Locker, a software system used across the USA & Canada for managing squash leagues, tournaments, etc. In particular, the doubles rating system is so unusable that Squash BC's official stance is to not use it for any decision making.

Problem Statement

We find the doubles rating system to be a particularly interesting problem. The idea is that you earn a rating as an individual, but you play within a partnership. However, the nature of squash means that your partners change quite frequently. This makes it difficult to answer questions such as:

• In a matchup of Alice & Bob vs Charlie & Diana, which team is likely to win?
• If Alice is a professional, Bob is a beginner, and Charlie & Diana are novice players?
• What if Charlie was also a beginner?
• What if everyone on the court is competent, except for one person?
• What if Alice & Bob play together all the time, but Charlie and Diana are first-time partners?
• How should ratings be impacted based on a match's outcome? Should both losing players have their ratings reduced by the same amount? Should both winning players have their ratings increased by the same amount? Does it change if it's a "friendly", "league", or "tournament" match.
• Can we predict an individual's rating? Or find individuals similar to them?
• How should teams be seeded in a tournament?

Opportunity

A benefit of Club Locker (and another system we used to use) is that we have a lot of match outcomes recorded and ready to be scraped off the website. There is a lot of interesting data available, and we think a lot of interesting questions to be answered. We could see a project that could involve the following "big data" aspects – data preparation, time series data, predictive modelling, anomaly detection, data visualization, & practical ML.

Collaboration

We have spoken with Squash BC's executive director, and he would be very interested in connecting with students on a project like this, perhaps with me as a liaison between SFU & Squash BC (as someone who understands the domain & the tech).