Stellar Classification

The classification scheme of galaxies, quasars, and stars is one of the most fundamental in astronomy. The goal of this project is to build a machine learning model that assists in classifying stars, galaxies and quasars (based on their spectral characteristics) with high accuracy. In this project, we used the Stellar Classification dataset from Kaggle. For features extraction, we analyzed the correlation between each feature, and eliminated the highly correlated and irrelevant features. Regarding model training, we explored three different algorithms for multi-class classification, including Random Forest, Logistic Regression and Naive Bayes. The performance of each model was fine-tuned and evaluated on the holdout set. We used accuracy, precision, recall and f1 metrics for evaluation, and utilized confusion matrix for performance visualization. In addition, we discussed the limitations of our work as well as future directions. The findings from this project will help better classify and analyze stellar observation data as more of the universe is observed.