Finding habitable exo planets using boosting algorithm
Abstract
The first moment man extended their boundaries outside of our Earth, from that moment
they were looking for another habitable planet, where they may live in future. Including
NASA, many international space organization already sent a number of satellite on this
mission. These mission have discovered thousands of new planetary candidates, many of
which have been confirmed through follow up observations. A primary goal of the mission is
to determine the occurrence rate of terrestrial-size planets within the Habitable Zone (HZ)
of their host stars. Though many approaches have been taken to confirm their habitability,
we tried a new approach by using boosting algorithms. We use the NASAs Extra Solar
planets dataset of 3,577 planets and use their various characteristics like their Mass, Radius,
Orbital Eccentricity, Temperature, Metallicity to determine the best set of alternatives of
Earth. We classified the dataset based on these variables and used Extreme Gradient Boosting
to compare the accuracy to find out our desirable results. We used different classifier to
ensure the best accuracy. So we used Ada Boosting Classifier, KNeighbor’s Nearest Classifier
(KNN), Gradient Boosting Classifier, Decision Tree Classifier and Random Forest Classifier
into our dataset.