📈 Linear Regression from Scratch

This project implements Linear Regression from scratch in Python, without relying on scikit-learn’s ready-made models.
It explores different optimization methods, compares their performance, and visualizes convergence and predictions.

👉 Open in Google Colab

---

🚀 Implemented Methods

• Closed-form solution (Normal Equation)
• Gradient Descent (GD)
• Stochastic Gradient Descent (SGD)
• Momentum-based Gradient Descent
• Adam Optimizer

---

📊 Features

• Custom StandardScaler for data normalization
• Metrics: MSE, RMSE, MAE, R² score
• Visualization utilities:
  • Loss curve comparison
  • Prediction vs. actual plots
  • Residual analysis
  • Convergence rate (log-scale)
  • Step-by-step gradient descent visualization

---

🖼️ Example Outputs

Loss Curves Comparison

Predictions Comparison

Gradient Descent Steps

---

⚙️ Getting Started

Clone the repo and open the notebook:

git clone https://github.com/yourusername/linear-regression-scratch.git
cd linear-regression-scratch
jupyter notebook LinearRegression.ipynb

Or run directly on Colab: 👉 Colab Link