An advanced machine learning platform for comprehensive race performance analysis, lap time prediction, and driver insights for the Toyota GR Cup Indianapolis Race 1.
RaceSense AI is a comprehensive racing intelligence platform providing predictions, sector insights, telemetry analysis, and improvement guidance for GR Cup drivers.
- Multi-Model Ensemble: 6 ML algorithms (XGBoost, Random Forest, Gradient Boosting, Ridge, LightGBM, CatBoost)
- High Accuracy: 0.513s Mean Absolute Error (0.51% prediction error)
- Per-Driver Optimization: Individual model selection for each driver based on performance
- Confidence Intervals: 90% prediction ranges with statistical validation
- Actionable Insights: 294 comprehensive recommendations with quantified time gains
- Sector Analysis: 3-sector lap breakdown with consistency scoring and comparative analysis
- Telemetry Processing: Real-time analysis of speed, throttle, brake, and RPM data
- Driver Profiling: AI-powered driving style classification and comparative performance metrics
- 5 Analysis Modules: Driver Overview, Lap Time Prediction, Driving Insights, Sector Breakdown, Leaderboard
- Real-Time Visualizations: Professional-grade charts with Plotly integration
- Full Vehicle Identification: Complete chassis tracking (e.g., GR86-049-88)
- Responsive Design: Optimized for desktop and mobile viewing
- Python 3.8 or higher
- 2GB RAM minimum
- 500MB available disk space
# Clone the repository
git clone https://github.com/Amethyst001/RaceSenseAI.git
cd RaceSenseAI
# Install required dependencies
pip install -r requirements.txt
# Download race data
# Visit: https://trddev.com/hackathon-2025/
# Download: indianapolis.zip
# Extract to project root (creates indianapolis/ folder)run_full_analysis.bat# Run complete analysis pipeline
python scripts\analyze_race.py
# Launch dashboard
cd dashboard
streamlit run dashboard.pyThe dashboard will be accessible at http://localhost:8502
Toyota/
├── scripts/ # Analysis scripts
│ └── analyze_race.py # Main analysis pipeline
├── run_full_analysis.bat # One-click launcher
├── requirements.txt # Python dependencies
├── dashboard/ # Streamlit dashboard
├── indianapolis/ # Race data (download separately)
├── outputs/ # Generated results
└── models/ # Trained ML models
Download race data from https://trddev.com/hackathon-2025/
File: indianapolis.zip
Extract to project root to create the following structure:
Toyota/
└── indianapolis/
└── indianapolis/
├── R1_indianapolis_BestLap.csv
├── R1_indianapolis_Laps.csv
├── R1_indianapolis_Results.csv
└── [other CSV files]
- Python 3.8+
- Windows (for batch files) or modify for Linux/Mac
- 2GB RAM minimum
- 500MB disk space
The system executes a comprehensive analysis workflow:
| Module | Description |
|---|---|
| Data Loading | Import and validate race data, calculate driver statistics |
| Telemetry Processing | Analyze speed, throttle, brake, RPM, and G-force data |
| Model Training | Train 6 ML algorithms per driver, select optimal model |
| Sector Analysis | 3-sector lap breakdown with consistency metrics |
| Insight Generation | Generate 294 actionable recommendations with time gains |
| Validation | Cross-reference predictions with official race results |
| AI Commentary | Intelligent driver profiling and comparative analysis |
- Performance metrics and comparative analysis
- Quick stats with prediction confidence
- Field ranking and gap analysis
- ML-powered next lap predictions
- 90% confidence intervals
- Model performance metrics
- Prediction quality dashboard
- 294 prioritized recommendations (HIGH/MEDIUM/LOW)
- GR86-specific technical guidance
- Priority distribution visualization
- Collapsible driver profiles
- 3-sector performance analysis
- Color-coded sector comparison (Amber/Cyan/Green)
- Strongest/weakest sector identification
- Improvement potential quantification
- Comprehensive driver rankings
- Best lap times and consistency scores
- Field-wide performance comparison
- Prediction Accuracy: 0.513s MAE (0.51% error rate)
- Data Coverage: 19 of 29 drivers analyzed (minimum 20 laps required)
- Insights Generated: 294 actionable recommendations
- Validation Score: Grade A- (90/100)
- Model Agreement: 6-model ensemble with per-driver optimization
Dashboard won't start:
pip install streamlit --upgrade
streamlit run dashboard/dashboard.pyMissing data:
Ensure indianapolis/ folder exists with CSV files from https://trddev.com/hackathon-2025/
Port conflict:
Edit dashboard.py or use: streamlit run dashboard.py --server.port 8503
dashboard/README.md- Dashboard documentationVALIDATION_REPORT.md- Logic validation results
- Python 3.8+: Primary development language
- Streamlit: Interactive dashboard framework
- Plotly: Professional data visualization
- Scikit-learn: Base ML algorithms and preprocessing
- XGBoost: Gradient boosting framework
- LightGBM: High-performance gradient boosting
- CatBoost: Categorical feature optimization
- Pandas & NumPy: Data manipulation and analysis
- 8 CSV Files: Race 1 telemetry, lap times, sector data, weather, official results
- 544 Telemetry Records: Comprehensive vehicle data
- 27 Drivers: Sector timing data coverage
This project was developed for the Toyota GR Cup Race Analytics challenge. Contributions, issues, and feature requests are welcome.
- Data Source: TRD (Toyota Racing Development) - https://trddev.com/hackathon-2025/
- Race Event: Toyota GR Cup Indianapolis Race 1
- Vehicle: GR86 Cup Car (spec racing series)
MIT License - See LICENSE file for details
For questions, issues, or collaboration inquiries:
- GitHub: @Amethyst001
- Repository: RaceSenseAI
v1.0.0 - Production Release (November 2025)
Built with ❤️ for motorsport analytics and data-driven racing performance optimization.