In this Major Qualifying Project (MQP), the team partnered with DraftKings Sportsbook to generate user-specific single bet and parlay predictions. To accomplish this, the team utilized clustering algorithms and evaluation methods to group users together. Multi-class classification algorithms including Linear Support Vector, Naïve Bayes, Stochastic Gradient Descent, and a neural network were then used to predict the most likely associations between bets/parlays and groups of users from the best clustering algorithm. The neural network classifier provided the best prediction performance and significantly outperformed random classification as well as classification to the most popular cluster. This project’s goal was to help DraftKings create a more personalized experience for their users by more accurately associating potential bets and parlays with each user.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Tomko, Cameron
• Fredo, Jack
• Dieselman, Ryan
• Lombardi, Jackson
• Casey, Grace

Publisher

• Worcester Polytechnic Institute

Identifier

• 105926
• E-project-042723-104248

Keyword

• Recommendation
• Clustering
• Sports
• Parlay
• Multi-class Classification
• Machine Learning

Advisor

• Brown, Donald R.

Year

• 2023

Sponsor

• DraftKings

Date created

• 2023-04-27

Resource type

• Major Qualifying Project

Major

• Data Science

Source

• E-project-042723-104248

Rights statement

• In Copyright

Last modified

• 2023-06-21