🏈 NFL Player Comparison API

A FastAPI backend that compares NFL players based on their most recent season stats using machine learning similarity metrics (cosine similarity).

🚀 Features

• Compare a given NFL player to others based on performance
• Uses real player data from nfl_data_py
• Standardized and normalized stats
• FastAPI-powered REST API

📦 Requirements

• Python 3.8+
• pip install -r requirements.txt

📁 Project Structure

nfl-player-comparer/
├── app/
│ ├── init.py
│ ├── main.py # FastAPI app
│ └── similarity_engine.py # ML logic and data loading
├── requirements.txt
├── README.md
└── .gitignore

⚙️ Getting Started

1. Clone the Repo

   git clone https://github.com/your-username/nfl-player-comparer.git
   cd nfl-player-comparer
   

2. Install Dependencies python3 -m venv venv source venv/bin/activate pip install -r requirements.txt

3. Run the Server uvicorn app.main:app --reload

4. Test the API Go to http://localhost:8000/docs for SwaggerUI

📈 Example Request

GET /similar?player=J.Jefferson&top_k=5

[
   "T. Hill",
   "A. Brown",
   "S. Diggs",
   "C. Lamb",
   "K. Allen"
]

📊 How It Works Loads current-year player data from nfl_data_py

Filters and standardizes stats (targets, receptions, yards, TDs)

Uses cosine similarity to compare players' performance profiles

Exposes a simple /similar API to find the closest matches

🔮 Future Ideas Add support for positions RBs, QBs, TEs, WRs

Add clustering with UMAP or PCA for visualization

Frontend for player selection and comparison graphs

Deploy to Railway, Render, or Hugging Face Spaces

📘 Credit nfl_data_py

scikit-learn

FastAPI