Identifying space threats using data-driven insights and predictive modeling.
Skillset Used : Astrophysics Data Analysis, Exploratory Data Analysis (EDA), Feature Engineering, Decision Trees, Machine Learning, Imbalanced Data Handling (Under & Oversampling)
🔍 What I did
- Analyzed near-Earth objects (NEOs) based on key features like diameter, velocity, and distance from Earth to detect potential space hazards.
- Integrated physics-guided insights with balanced under and oversampling techniques for better Exploratory Data Analysis (EDA).
- Developed a predictive model using decision trees, achieving 96% accuracy in identifying hazardous objects.
- Optimized feature selection and model tuning to enhance precision and reduce false positives.
📈 Impact & Insights
- Improved hazard detection, helping scientists assess the potential risks of asteroids and comets.
- Bridged physics and data science, leveraging both theoretical and machine learning-driven approaches.
- Enhanced classification accuracy, making predictions more reliable for space monitoring agencies.
- Laid the groundwork for further AI-driven space research and planetary defense strategies.
🚀 Learning Outcomes
- Strengthened expertise in space data analysis & predictive modeling.
- Gained hands-on experience in handling imbalanced datasets with under/oversampling techniques.
- Improved decision-making using physics-based insights combined with machine learning.
- Explored the intersection of AI, astrophysics, and planetary defense.