Exploring XGBoost as an Effective Machine Learning Algorithm for Stellar Spectral Data Classification in Astronomy

The discipline of astronomy has experienced a substantial metamorphosis due to the advent of contemporary telescopes, which facilitate the accumulation of extensive spectral information. The processing of data through traditional manual means has become unfeasible owing to its vast volume, thereby necessitating the adoption of automated classification. The advent of machine learning algorithms and artificial intelligence has enabled the efficient and accurate processing of spectral data, thereby facilitating a more comprehensive analysis of celestial objects. Prior research has investigated the efficacy of diverse machine learning algorithms in the context of stellar classification endeavors, as well as the significance of pre-processing spectral data. The present investigation centers on the utilization of XGBoost, a machine learning algorithm that exhibits the ability to handle heterogeneous data types and address imbalanced datasets, thus rendering it a viable option for classification endeavors. In this research, we found that Random Forest and XGBoost gave good result compared to other classifier. In contrast, reduction dimension using PCA are lowering the achieved result.