NumPy Linear Regression from Scratch 🚀

A pure NumPy implementation of Linear Regression to understand ML fundamentals.

📊 Results

• R² Score: 0.9459 (94.59% accuracy)
• Final Loss (MSE): 673,399,536.67
• Dataset: 100 house samples (size, bedrooms → price)

🚀 Features

• ✅ Pure NumPy implementation (no sklearn)
• ✅ Gradient descent optimization from scratch
• ✅ Feature normalization
• ✅ MSE loss function
• ✅ Visualization of results

📦 Installation

pip install numpy matplotlib

💻 Usage

python projects/main.py

🧮 How It Works

1. Forward Pass: y = X @ weights + bias
2. Calculate Loss: MSE = mean((y_pred - y_actual)²)
3. Compute Gradients: Derivatives of loss
4. Update Parameters: weights -= learning_rate * gradient
5. Repeat until convergence

📚 What I Learned

• NumPy array operations and matrix multiplication
• Gradient descent algorithm
• Why we square errors in loss functions
• Feature normalization importance
• Model evaluation with R² score

👨‍💻 Author

Built while learning machine learning fundamentals from scratch!

📝 License

MIT License - feel free to use for learning!