Analyzing Astronomical Data with Machine Learning Techniques

TL;DR

This paper reviews popular machine learning classification algorithms and applies them to astronomical data, specifically classifying nonvariable and RR Lyrae stars from SDSS survey, evaluating their accuracy and F1-score.

Contribution

It introduces the application of multiple ML classification models to astronomical star data and compares their performance on real observational datasets.

Findings

Decision tree and random forest achieved highest accuracy.

Support Vector Machine showed strong F1-score performance.

Machine learning models significantly improve classification efficiency in astronomy.

Abstract

Classification is a popular task in the field of Machine Learning (ML) and Artificial Intelligence (AI), and it happens when outputs are categorical variables. There are a wide variety of models that attempts to draw some conclusions from observed values, so classification algorithms predict categorical class labels and use them in classifying new data. Popular classification models including logistic regression, decision tree, random forest, Support Vector Machine (SVM), multilayer perceptron, Naive Bayes, and neural networks have proven to be efficient and accurate applied to many industrial and scientific problems. Particularly, the application of ML to astronomy has shown to be very useful for classification, clustering, and data cleaning. It is because after learning computers, these tasks can be done automatically by them in a more precise and more rapid way than human operators. In view of this, in this paper, we will review some of these popular classification algorithms, and then we apply some of them to the observational data of nonvariable and the RR Lyrae variable stars that come from the SDSS survey. For the sake of comparison, we calculate the accuracy and F1-score of the applied models.