NFL Game Predictor

Welcome to the NFL Game Predictor repository! This project uses a Random Forest Classifier to predict the outcomes of NFL games based on team statistics specifically for only the 2022-2023 Season. The predictions include confidence scores and take into account home-field advantage. This repository contains the backend machine learning model and all necessary datasets. Our front-end is hosted on Streamlit at NFL Game Predictor App.

---

Repository Structure

Files and Directories

• Random_forest_FINAL.ipynb: The main Jupyter Notebook containing the machine learning pipeline, from data preprocessing to model training and evaluation.
• Datasets:
  • Individual team data files (e.g., 49ers_data.csv, bears_data.csv): Contain team-specific stats used to train and test the model.
  • total_schedule.csv: Consolidated schedule of games.
  • football_stats.csv: Includes aggregated and additional metrics for analysis.
• Model Files:
  • random_forest_model.joblib: Trained Random Forest model.
  • scaler.pkl: Scaler used for feature normalization.
• model_webpage.py: The Python script for deploying the model via Streamlit.
• requirements.txt: Contains all the Python dependencies required to run the project.
• README.md: This readme file.

---

Installation and Setup

1. Clone the Repository:

   git clone https://github.com/your-username/nfl-game-predictor.git
   cd nfl-game-predictor

2. Install Dependencies: Use the requirements.txt file to install all necessary libraries:

   pip install -r requirements.txt

3. Run the Streamlit App: Start the Streamlit front end to visualize predictions:

   streamlit run model_webpage.py

4. Data and Model Files: Ensure all dataset files and model files (random_forest_model.joblib and scaler.pkl) are in the correct directory.

---

Features

• Game Prediction: Predict the winner of a matchup with a confidence score. Includes home-field advantage as a feature.

• Data Visualization: Includes exploratory data analysis (EDA) visualizations, such as feature distributions, correlation matrices, and boxplots.

• Streamlit Integration: A user-friendly interface for entering team matchups and viewing predictions.

---

How It Works

1. Data Preprocessing:

   • Team-specific data is preprocessed, with features like TotalYards, PassYards, RushYards, and Turnovers used for training.
   • Home-field advantage is included as a binary feature.

2. Model Training:

   • A Random Forest Classifier is trained using these features, and hyperparameters are optimized for better accuracy.

3. Prediction:

   • Users input two teams and specify the home team.
   • The model predicts the winner and provides a confidence score for the prediction.

---

Example

To illustrate the model’s performance, we predicted the matchup between the Lions and Commanders. The model correctly predicted the Lions to win with a confidence score of 69%, aligning with the Lions’ strong performance this season and the Commanders’ recent struggles.

---

Future Work

• Incorporating player-level data and recent performance metrics.
• Expanding model features to include weather conditions, injuries, and advanced analytics.
• Enhancing the user interface for a more interactive experience.

---

Authors

• Akhil Sharma
• Bhavesha Sasikumar
• Jackson Cmelak
• Denil Neil