Linear Regression from Scratch

This project demonstrates a complete implementation of Linear Regression using Python and NumPy, built from scratch with gradient descent optimization. It also includes a comparison with the LinearRegression model from the scikit-learn library to validate the implementation.

---

Project Description

Linear Regression from Scratch

• Implemented a simple linear regression model to fit a line to the dataset of Years of Experience vs Salary.
• Used Mean Squared Error (MSE) as the loss function.
• Optimized the slope (m) and intercept (c) parameters using gradient descent.
• Training initialized with random values for m and c, a learning rate of 0.01, and trained over 5000 epochs.
• Loss is printed every 5000 epochs to monitor training progress.
• Final regression line is plotted against the data points.

Linear Regression Using Scikit-learn

• Trained the built-in LinearRegression model from scikit-learn on the same dataset.
• Compared the intercept and slope learned by scikit-learn with those from the scratch implementation.
• Plotted both regression lines for visual comparison.
• Calculated and printed the MSE loss for the scikit-learn model.

Comparison

• Visual plot showing the scratch implementation’s regression line and the scikit-learn regression line together with the original data points.

---

Dataset

The dataset used is Salary_Data.csv with two columns:

• YearsExperience: Number of years of work experience
• Salary: Corresponding salary

---

How to Run

1. Clone this repository:

   git clone https://github.com/CodeLoopdroid/Linear-Regression-from-Scratch.git
   cd Linear-Regression-from-Scratch