Schedulion

Date of Completion

5-4-2022

Degree Type

Honors Thesis

Discipline

Computer Science (CMSI)

First Advisor

Andrew Forney

Abstract

Schedulion is a web application designed to aid the Loyola Marymount basketball team with schedule building. To determine a matchup's efficacy, the application focuses on the NET score and winning percentage. NCAA basketball teams are ranked based on their NET score, which is determined by a formula unknown to the public. In order to get the statistics necessary for our application, we pull data from kenpom, which is currently the best source for college basketball statistics. Our website uses the help of kenpompy, a tool our team helped develop, to achieve this end. The application uses a linear regression model to determine the NET score, which yielded 98.1% accuracy on testing data after training. The application then uses a recurrent neural network model in order to determine LMU's predicted winning percentage. The winning percentage model yielded a 77.8% accuracy on testing data after training. The application has a firebase, flask, and reach tech stack. Firebase enables the application to store NET score and winning percentage values in a database once they have been calculated to increase the latency of our application. Firebase Admin commands allow us full CRUD (create, read, update, delete) capability for schedule and game objects. The code for schedulion can be accessed at https://github.com/nmorgan8/schedulion, and the deployed application can be found at https://schedulion.vercel.app/.

Link to Full Text

Share

COinS