Detection and classification of exoplanets using hybrid kNN model

As you can imagine, a planet itself does not emit light, but the stars orbiting the planet do. If this star is observed over a long period of time, there may be regular “enfeebling” of the flux (light intensity). This is evidence that there may be objects in orbit around the star. Such stars can be considered a “candidate system”. For example, further investigation of candidate systems with satellites that capture light of different wavelengths may solidify the assumption that candidates can actually be “confirmed.” Planet identification was usually performed only by a team of astronomers and astrophysicists, using methods and tools available only to those with academic training. NASA's Extrasolar Planetary Exploration Program has introduced multiple state-of-the-art satellites that can collect a variety of data about celestial bodies of interest and assist in exploring those celestial bodies. Now that satellite data is easily accessible, data analysts can assist in planet detection. Our study uses several classification models and datasets to assign the possibility that the observations are exoplanets. Taking into account the mass and orbital period of the planets, K-means clustering method was used to classify three large groups of exoplanets, Hot Jupiter, long-period giants, and other planets.