🚀 Asteroids Classification using Machine Learning

This project aims to classify asteroids using a machine learning model based on their physical and orbital characteristics. The dataset is cleaned, preprocessed, and analyzed to extract meaningful patterns that improve classification accuracy.

📌 Project Overview

• Loads and preprocesses asteroid data, removing redundant and irrelevant features.
• Encodes categorical features like NEO (Near-Earth Object) and PHA (Potentially Hazardous Asteroid).
• Implements feature engineering to refine data representation.
• Trains a classification model to predict asteroid types.
• Evaluates model performance using various metrics.

---

🛠️ Technologies Used

• Python (Primary language)
• Scikit-learn (Machine Learning models)
• Pandas & NumPy (Data handling)
• Matplotlib & Seaborn (Data visualization)
• Jupyter Notebook (Exploratory data analysis)

---

🔬 Data Preprocessing Steps

1️⃣ Feature Selection

• Removed unnecessary attributes like id, name, and redundant time representations (epoch_mjd, epoch_cal).

2️⃣ Feature Cleaning

• Eliminated missing values and irrelevant attributes.
• Unified redundant features (per, per_y) that had different scales.

3️⃣ Feature Engineering

• Encoded categorical variables (neo, pha, class) into numerical representations.
• Computed missing neo values using perihelion distance (q).

4️⃣ Final Dataset Preparation

• Ensured feature consistency and normalization for model training.

---

🤖 Model Implementation

• Applied Supervised Learning techniques to classify asteroids.
• Split dataset into training and testing sets.
• Evaluated model performance using accuracy, precision, recall, and F1-score.

---

📊 Results & Insights

• Successfully classified asteroids based on orbital and physical properties.
• Encoding neo based on perihelion distance (q) improved classification accuracy.
• Some features were redundant and removing them enhanced model efficiency.