We are a group of 2 Computer Science students and a Mechanical Engineering student who are all passionate about college and professional athletics. When deciding on a project, we wanted to look into how to compare teams across historical boundaries as that is a common source of arguments. Two of us are currently enrolled in EECS 738, so we wanted to develop a product related to what we were currently learning in class. As a result, we settled on analyzing statistics generated by kenpom, a famous basketball statistician, and create our own data metrics based on similarities we found.

What it does

This project is a demonstration of direct implementations of different machine learning algorithms in an attempt to create a classification model on NCAA Men's basketball games from 2002 to 2020. With this data model, we are able to compare teams across different years to provide accurate head to head matchups. We have demonstrated this by allowing users to select every Division I Men's basketball team over the past two decades and pit them against any other team in the dataset. In addition, we have created a bracket checking system which allows the user to select any NCAA Tournament region and simulate our model against the real and historical outcomes.

How we built it

Mainly using data analytics tools provided by pandas / SciKit Learn and the front-end was primarily developed in javascript utilizing the react library.

Challenges we ran into

When developing our project we ran into relatively few issues, but once it came time to focus on deployment there were quite a few hiccups. Mainly it was based around the fact that we were all relatively inexperienced with having to implement a reverse proxy system.

Accomplishments that we're proud of

Being able to smooth out large amounts of noisy data in a relatively short amount of time.

What's next for The Ides of March

Carrying this project onto EECS 738 at KU.

Share this project: