|
|
 |
| |
|
|
|
Machine Learning Approaches for Galaxy Categorization |
|
|
|
PP: 139-146 |
|
|
|
Author(s) |
|
|
|
Mohamed Ben Hassen,
Kaies debbabi,
|
|
|
|
Abstract |
|
|
| The main topic of this study is the categorization of galaxies using machine learning methods on the Stellar categorization - SDSS17 dataset. The collection consists of 100,000 observations obtained using the Sloan Digital Sky Survey (SDSS). Each observation is identified as a star, galaxy, or quasar by one class column and 17 feature columns. Three classification models are used in the study: Decision Tree, XGBoost, and K-Nearest Neighbors (KNN). Exploratory data analysis provides insights into the distribution and correlations of the data after preprocessing of the dataset, which includes converting categorical class values to numeric values and removing unnecessary features. The models are then trained on the data, and different metrics such us accuracy, F1 score, confusion matrix, and classification ratio were used to assess how well the models performed. Based on the accuracy score of 97.5%, XGBoost performs better than the other models, with KNN and Decision Tree coming in second and third, respectively. Results obtained This work offers insightful information on the categorization of galaxies by machine learning methods, with possible uses in data analysis and astronomical research.
|
|
|
|
|
|
|
|
