NASA Asteroids Classification

Machine learning project for classifying whether near-Earth asteroids are potentially hazardous using NASA observational data.

Dataset

Source:
https://www.kaggle.com/datasets/lovishbansal123/nasa-asteroids-classification

• Total samples: 4687
• Features: Orbital and physical characteristics
• Target variable: Hazardous

---

Project Workflow

1. Data cleaning
2. Correlation analysis
3. Feature selection
4. Data standardization
5. PCA dimensionality reduction
6. Model training
7. Hyperparameter tuning
8. Model evaluation

---

Models Used

Decision Tree

• Test Accuracy: 87.74%

Logistic Regression

• Test Accuracy: 95.84%

Random Forest

• Test Accuracy: 94.03%