Inspiration After we started exploring the dataset, what really inspired us was how much astronaut safety comes down to details that often get overlooked: age, time in space, and past missions. The more we looked in, the more we realized there was a real opportunity to use data science to highlight risks before they became problems. Space travel is already risky, so building something that could make missions even a little safer felt truly meaningful.
What it does
- Predicts mission duration for astronauts.
- Generates a risk score with confidence intervals.
- Explains risk factors and provides recommendations.
- Displays predictions and analytics through an interactive React dashboard.
How we built it
- Backend: Flask + Python REST API.
- Model: Stacked ensemble of Random Forest, Gradient Boosting, Neural Net, and SVR.
- Data: Cleaned, engineered features (age groups, experience levels), validated inputs.
- Frontend: React components for predictions, visualizations, and system health.
Challenges we ran into
- Cleaning data and handling outliers.
- Tuning ensemble so it beat individual models.
- Debugging frontend and backend communication.
- Managing hackathon time pressure while building end-to-end.
Accomplishments that we're proud of
- Built a working full-stack system in limited time.
- Developed a stacked ensemble model with strong results.
- Created a React dashboard that shows outputs clearly.
- Added validation/error handling to manage bad inputs.
- Collaborated effectively across backend, frontend, and ML.
What we learned
- How to use an Stack Ensemble Model
- Importance of time in Space Missions
What's next for Risk Prediction for Space Missions
- Add more astronaut data (health, psychological, mission complexity).
- Deploy to cloud with Docker for real-world usability.
- Add explainable AI tools to show why risks are flagged.
- Improve dashboard with more analytics and mobile support.
Log in or sign up for Devpost to join the conversation.