Ft Linear Regression

📄 Overview

The goal of this project is to introduce basic machine learning concepts. I created a program to predict car prices using a simple linear regression model trained with a gradient descent algorithm. The project focuses on a specific example—predicting car price based on mileage—but once complete, you can apply the model to any similar dataset.

You’ll build two programs. The first will predict car prices: it prompts you for a mileage value, then returns the estimated price for that mileage. The second program trains the model by reading a dataset and applying linear regression to adjust the model for accurate predictions.

📦 Installation

Ensure you have Python 3.10 or newer. Download the latest version of Python from the official Python website.

1. Clone the Repository:

   git clone https://github.com/DevJ2K/ft_linear_regression.git
   cd ft_linear_regression

2. (Optional) Create a Virtual Environment:

   It’s recommended to use a virtual environment for managing dependencies.

   python3 -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install Dependencies:

   If dependencies are listed in requirements.txt, install them with:

   pip install -r requirements.txt

🔍 Usage

1. Train the Model:

   Run the training script:

   python3 train_model.py -h

   This should display:

   usage: train_model.py [-h] -f FILE [-i ITERATIONS] [-lr LEARNING_RATE] [-a]
   
   Train a model using the provided configuration.
   
   options:
     -h, --help            show this help message and exit
     -f FILE, --file FILE  Configuration file for training the model.
     -i ITERATIONS, --iterations ITERATIONS
                           Number of iterations for model training (default: 1000).
     -lr LEARNING_RATE, --learning-rate LEARNING_RATE
                           Learning rate influencing model convergence speed (default: 0.01).
     -a, --animate         Enable training curve animation.

2. Predict Using the Model:

   To make predictions based on mileage:

   python3 predict.py -h

   Expected output:

   usage: predict.py [-h] -f FILE -m MILEAGE [-s] [-g]
   
   Use a pre-trained model to make predictions based on input values.
   
   options:
     -h, --help            show this help message and exit
     -f FILE, --file FILE  File containing the trained model's parameters.
     -m MILEAGE, --mileage MILEAGE
                           Mileage for price prediction.
     -s, --steps           Display detailed prediction steps.
     -g, --graph           Visualize prediction results with a graph.

🚀 Examples

Training the Model

python3 train_model.py -f data.json -i 1500 -lr 0.01

Expected output:

SUCCESS: Training data saved in '/$REPOSITORY/ft_linear_regression/models/model_data.json'.

** STATISTICS ***********************************
WARNING: Theta trained on standardized data.
Thetaθ(0) (Weight): [-0.85613918]
Thetaθ(1) (Bias): [2.38235357e-16]
Thetaθ - Matrix form:
[[-8.56139178e-01]
    [ 2.38235357e-16]]

** LEARNING INFORMATIONS ************************
Iterations : 1500
Learning Rate : 0.01
Coefficient of determination : 0.7329747078314375

** STANDARDIZATION INFORMATIONS *****************
Mean X (μx): 101066.25
Mean Y (μy): 6331.833333333333
Standard Deviation X (σx): 51565.1899106445
Standard Deviation Y (σy): 1291.8688873961714

Example graph output:

Using the Model for Prediction

python3 predict.py -f model_data.json -m 170000 -s -g

Expected output:

** STATISTICS ***********************************
WARNING: Theta trained on standardized data.
Thetaθ(0) (Weight): [-0.85613918]
Thetaθ(1) (Bias): [2.38235357e-16]
Thetaθ - Matrix form:
[[-8.56139178e-01]
    [ 2.38235357e-16]]

** LEARNING INFORMATIONS ************************
Iterations : 1500
Learning Rate : 0.01
Coefficient of determination : 0.7329747078314375

** STANDARDIZATION INFORMATIONS *****************
Mean X (μx): 101066.25
Mean Y (μy): 6331.833333333333
Standard Deviation X (σx): 51565.1899106445
Standard Deviation Y (σy): 1291.8688873961714

** 1. STANDARDIZED INPUT ************************
- Standardize input: standardize(x) = (x - μx)/σx
- standardize(170000) = 1.336827230142134

** 2. MATRIX TRANSFORMATION OF INPUT *************
- Transform standardized input to matrix format.
- 1.336827230142134 → [[1.33682723 1.        ]]

** 3. APPLY MODEL ON MATRIX *********************
- Use theta (θ) for matrix product and provide prediction.
- model(X) = X.θ

- model([[1.33682723 1.        ]]) = -1.1445101658636485

** 4. DESTANDARDIZED PREDICTION *****************
- Destandardize for interpretation.
- destandardize(y) = y * μy + σy
- -1.1445101658636485 → 4853.276258745454

PREDICTION → Estimated price for mileage of 170000km: 4853.28€.

Example graph output: