MISSION OVERVIEW

Developed for the NASA International Space Apps Challenge 2025, this project tackles the "A World Away: Hunting for Exoplanets with AI" challenge. The objective is to automate the validation of planetary candidates from the Kepler K2 mission data.

NEURAL ARCHITECTURE

The system utilizes a dual-engine approach for maximum classification precision:

• XGBOOST ENGINE: A gradient-boosted decision tree architecture optimized for high-dimensional feature spaces. Currently the primary engine.
• D-TREE ENGINE: A secondary classification layer used for redundancy and decision path transparency.

DATA SOURCES

All training data is derived from the NASA Exoplanet Archive, specifically the K2 Cumulative Table. Features were engineered to represent stellar temperature variance, planetary radius relative to host star, and orbital stability metrics.