Machine Learning Method for Predicting the Merge and Morphology of Galaxies through Near-Infrared Spectroscopy

Astronomy is experiencing rapid growth in the size and complexity of data. This reinforces the development of data-driven science as a useful complement to the current model of model-based data analysis. In spite of this, traditional merger analysis of catalogs is mostly done through visual inspection by trained experts. A method that is not efficient today, because the subjectivity of manual classification has made the result of the process very dependent on the intuition of the analyst and the type and quality of the image. Hence, this study focuses on data processing methods based on Artificial Intelligence (AI) algorithms and investigates the possibility of a pattern among astronomical data to predict the merger of galaxies. The modeling is done in two phases. The first phase deals with the classification between minority and majority classes and the second phase search for any distinction between minority classes. In both phases, various algorithms such as Naive Bayes, Random Forest, and Generalized linear model (GLM) and Neural network are used to ensure the best results according to the research data. The best results for both phases obtained from the implementation of the GLM algorithm with the accuracy of 70.28 % and 76.51% for the first and second phase respectively.