Griffin, Ermina, Yaphet, and Aidan's capstone project for the Certificate of Data Science at the Georgetown University School of Continuing Studies (SCS) (Cohort 23, Spring - Summer 2021).
Photo courtesy of Medium user Brian Bockleman
“Thank you... fantasy football draft, for letting me know that even in my fantasies, I am bad at sports.”
Jimmy Fallon
Fantasy football allows fans to act as team managers by drafting, trading for, acquiring, and playing real football players on fantasy football platforms, scoring points using a scoring system based on real life performance of their players. Fantasy football platforms (such as ESPN or Yahoo!) apply their own analytics to project player performance weekly during the NFL season in preparation for the upcoming week. It is not uncommon for players to score well above or far below their platform-projected fantasy score, leaving fantasy managers wondering which players to draft, trade for, acquire, and play.
We want to know what influences a player's actual fantasy score so that we can make data-driven decisions when building a team or weekly starting lineup during the NFL season. To do so, we're using 2019 and 2020 NFL season data and ESPN fantasy football projection data for this project.
For a full explanation of our project from beginning to end, see our final presentation and read our final paper
| Directory | Contents |
|---|---|
| bin | Python data ingestion and database setup scripts |
| deliverables | Georgetown SCS-required documents for project reporting |
| fixtures | Raw and cleaned data, images, and database |
| foo | Final machine learning models (global, offense, defense, and wide receiver) |
| notebooks | Cleaning, Exploratory Data Analysis, Machine Learning Jupyter Notebooks |
| tests | ✨ Future home of automated testing scripts |
First, clone this repository by entering the following into terminal (Mac) or powershell (Windows):
git clone https://github.com/georgetown-analytics/Cloudy-with-a-Chance-of-Football.git
Next, switch to this directory by entering the following into terminal or powershell:
cd Cloudy-with-a-Chance-of-Football
Finally ensure you have the proper package versions downloaded by entering the following line into terminal or powershell:
pip install requirements.txt
At this time, we do not have guidance on how to contribute to this project.
Special thanks to Kaggle users mur418 and tobycrabtree, Fantasy Data, NFLSavant, Data Pros, and NFL Weather for data use.