A brief primer on exoplanet science and the role of explainable machine learning in modern astronomy.
An exoplanet is any planet orbiting a star outside our solar system. Since the first confirmed detection in 1992, more than 5,000 exoplanets have been catalogued, ranging from small rocky worlds to gas giants larger than Jupiter.
When a planet passes in front of its host star, it blocks a tiny fraction of light. Space telescopes like Kepler and TESS record these periodic dips in brightness. The depth, duration, and periodicity reveal the planet's size and orbit.
Astronomers need to trust and verify each classification. SHAP explanations reveal which stellar features drove a prediction, turning a black-box model into a research-grade decision support tool.
Our pipeline ingests data from the NASA Exoplanet Archive — an open catalogue of confirmed exoplanets and candidate observations from missions like Kepler, K2, and TESS.