We model exoplanets as nodes in a graph, connecting them using Euclidean similarity based on planetary and stellar features. Our goal is to identify Earth-like planets and use inferential statistics to analyze whether stellar characteristics are related to the likelihood of hosting such planets.