F1 Race Prediction Project

Machine learning models to predict Formula 1 race outcomes using historical data and driver statistics.

Project Overview

This project implements multiple machine learning algorithms to predict Formula 1 race winners using data from 1950-2020. Models include Random Forest, SVM, RNN, XGBoost, and Logistic Regression.

Dataset

• Source: Formula 1 World Championship (1950-2020)
• Features include:
  • Driver Elo ratings
  • Grid positions
  • Circuit characteristics
  • Historical performance

Models Implemented

• Random Forest (97.40% accuracy)
• Support Vector Machine (96.80% accuracy)
• Recurrent Neural Network (96.19% accuracy)
• XGBoost (96.08% accuracy)
• Logistic Regression (95.80% accuracy)

Project Structure

F1-Prediction/
├── data/
│   ├── circuits.csv
│   ├── drivers.csv
│   ├── races.csv
│   ├── results.csv
│   └── driver_elo.csv
├── models/
│   ├── rnn.py
│   ├── xgboost_model.py
│   ├── svm_model.py
│   ├── random_forest.py
│   └── logistic_regression.py
└── simulate/
    └── simulateRNN2024.py

Setup and Installation

1. Clone the repository:

git clone https://github.com/yourusername/F1-Prediction.git
cd F1-Prediction

2. Install requirements:

pip install -r requirements.txt

3. Train models:

python rnn.py
python xgboost_model.py
python svm_model.py
python random_forest.py
python logistic_regression.py

4. Run predictions:

python simulateRNN2024.py

Results

Model predictions are saved in:

• data/predicted_models/predicted_results2024RNN.csv
• data/predicted_models/predicted_results2024XGB.csv
• data/predicted_models/predicted_results2024SVM.csv
• data/predicted_models/predicted_results2024RF.csv
• data/predicted_models/predicted_results2024LR.csv

Contributors

• Ryan Shafi
• Arnab Nath

License

This project is licensed under the MIT License - see the LICENSE file for details.