One-Variable-Linear-Regression

One Variable Linear Regression from Scratch using gradient descent and using scikit-learn

Overview

This project demonstrates linear regression, gradient descent and squared error function from scratch and also using scikit learn for single variable. This project uses a small dataset of student test scores and the amount of hours they studied.

Gradient descent visualization

Sum of squared distances formula (to calculate our error)

Partial derivative with respect to b and m (to perform gradient descent)

Dependencies

• numpy

Use pip to install numpy to install numpy. Works with Python 2.x and 3.x

Execution

To execute Linear Regression from Scratch, just run the file using Python

python "Linear Regression from Scratch.py"

The output will look like this

Starting with m as 0 and b as 0 we get error 2782.5539172416056
Running....
After 1000 iterations, m as 1.472003357146311 and b as 0.2658134872230222 we get error 56.261600442221535

To execute Linear Regression using Scikit Learn, just run the file using Python

python "Linear Regression using Scikit Learn.py"

The output will look like this

We get m as 1.3381037220246799 and b as 7.221214607193161.
Mean squared error - 137.14657579981292.