HYPOTHESIS

We hypothesized that we could train a machine learning model to predict teams win/loss records for future seasons evaluating team stats from past seasons.

PROCESS
  • Pull data from pro-football-reference.com and the sports-reference.com API
  • Process the data using Pandas
  • Create a number of regression models and graph the results via matplotlib
  • Create a model to predict current season’s W/L record
  • Use that model to predict future season’s W/L Record
  • Determine the viability of our models by comparing results
  • Publish our findings to GitHub website
alternative
MODEL

Shows ML Model HTML page

alternative
FINDINGS

We learned that current season data doesn’t provide a solid base to predict future seasons due to the many intangible factors that happen between seasons. Our models came up with R-squared numbers of .13 and lower.

alternative
ABOUT US

Combining Our Passion for Data Science, Machine Learning, and Sports

Jeremy Brent- Data Scientist. Clark University, graduated 2019. Major: Biology & Psychology. LinkedIn

Matt Sadowski- Senior Data Analyst at SHI International. LinkedIn

Brian Remite- 12 years of experience as an SQL Database Analyst. LinkedIn

Bryan Wilson- Data Scientist/Web Developer. Kean University, graduated 2017. Major: Business Administration, concentration in Finance. LinkedIn

Alim Memon- Data Scientist & senior at Rutgers University. Major: Information Technology and Informatics; Interested in Machine Learning and AI. LinkedIn

  • Rutgers Data Science Cohort 2020

Copyright © 2020 GitHub Repository