woolimi / ft_linear_regression

ft_linear_regression

Overview

ft_linear_regression is a 42 machine learning project to create a program that predicts the price of a car by using a linear function train with a gradient descent algorithm. Dataset consists of a pair of price and mileage in src/data.csv. In this project we are free to use any language, so I used Javascript(NodeJS and React for visualization)

How to use

1. Install package

   > pnpm i
   

2. Prediction before train

   Try to predict price before training our src/model.json

   > pnpm predict
   > milage: 200000
   > price: 0
   

3. Train

   > pnpm train
   ===========================
   Theta0 :  8008.400720135726
   Theta1 :  -0.02144848806185557
   ===========================
   

4. Prediction after train

   > pnpm predict
   > milage: 200000
   > price: 3718.7031077646125
   

5. Visualization (React App localhost:8080)

   > pnpm bonus
   

   By changing the learning rate, you can observe how the gradient descent algorithm is performed.