Supervised Classification for Analysis and Detection of Potentially Hazardous Asteroid

Abstract

The use of Artificial Intelligence (AI) in solving real- time problems are increasing day by day with the increase in the availability of data and computation power. It is now substantial to use AI-based tools and techniques in space science. Asteroids, rocky objects that orbit around the sun, often produce an array of effects that cause harm to humans and biodiversity on earth. Such effects can cause wind blast, overpressure shock, thermal radiation, cratering, seismic shaking, ejecta deposition, tsunami, and many more. With the availability of data on asteroid parameters and nature, it provides an opportunity to use Machine Learning (ML) to address this problem and reduce the risk. This paper presents a thorough study on the impact of Potentially Hazardous Asteroids (PHAs) and proposes a supervised machine learning method to detect whether an asteroid with specific parameters is hazardous or not. We compare manifold classification algorithms that were implemented on the data. Random forest gave the best performance in terms of accuracy (99.99%) and average F1- score (99.22%).