🌌 Yuri Night: Asteroid Classification Model

🎉 A machine learning project to honor space exploration and planetary defense.

---

🛰️ Classification Overview

This repository contains an ML pipeline for classifying asteroids into:

• 🪐 NEO (0) – Near-Earth Objects
• ☄️ PHA (1) – Potentially Hazardous Asteroids
• 🌍 NHA (2) – Non-Hazardous Asteroids

Developed to celebrate Yuri’s Night — the "World Space Party" 🌍🚀

---

🧠 What This Project Does

✅ Classifies asteroids using orbital and photometric data
✅ Uses ML models like KNN, SVM, Random Forest, and XGBoost
✅ Tackles class imbalance using SMOTE
✅ Adds meaningful features like MOID to SMA Ratio
✅ Visualizes patterns and skewness in the data

---

📊 Dataset Overview

Asteroid data includes:

• H – Absolute Magnitude
• a – Semi-Major Axis
• q – Perihelion Distance
• moid – Earth MOID
• diameter, albedo, e, i, etc.

Classification logic:

• NEO: q < 1.3 AU
• PHA: moid ≤ 0.05 AU and H ≤ 22.0
• NHA: All others

📁 Dataset Source: NASA CNEOS NEO Dataset

---

🤖 ML Models Used

• K-Nearest Neighbors
• Support Vector Machine
• Random Forest
• XGBoost Classifier

---

⚖️ Handling Class Imbalance

Initial class distribution was skewed.
Applied SMOTE to balance class ratios manually to:

• 🪐 NHA (2): 30%
• ☄️ PHA (1): 10%
• 🌍 NEO (0): 10%

🔗 Learn about SMOTE

---

📈 Visualizations

• 📊 Heatmaps (Asteroid type vs Year)
• 📉 Histograms for skewness
• 📌 Correlation matrices
• 📊 Class bar plots

Built with Seaborn and Matplotlib

---

📂 Project Structure

├── data/ ├── notebooks/ ├── src/ ├── visualizations/ └── README.md

---

🚀 Run Locally

# Clone the repo
git clone https://github.com/yourusername/yuri-night-asteroid-classifier.git

# Install dependencies
pip install -r requirements.txt

# Open Jupyter or run in Colab


---

Let me know if you want this saved in a `.md` file — or if you want dark/light mode badge support too.