Technical Projects
98.48% Accuracy
Machine Learning
Predicting Stars, Galaxies & Quasars
Supervised ML Pipeline for SDSS Photometric Classification
Building a supervised machine learning pipeline to classify astronomical sources from the Sloan Digital Sky Survey (SDSS) based on photometric features. This simulates a real data-intensive astronomy workflow essential for modern large-scale surveys.
Methodology
- Data cleaning (dropping objid/specobjid)
- Feature scaling with StandardScaler
- Hyperparameter tuning via GridSearchCV
Models Evaluated
- Random Forest (98.48% Accuracy)
- XGBoost & Neural Networks (MLP)
- Logistic Regression & SVM
Tools Used
- Pandas, NumPy, Matplotlib, Seaborn
- Scikit-Learn & Streamlit
- Publicly available SDSS CSV dataset