Welcome, ashen one. You've stumbled upon the ft_linear_regression project — the saga of one lonely Python script trying to predict used car prices based solely on how far they’ve been driven. No sci-fi libraries. No black-box sorcery. Just shadow math, madness, and mileage.

"Life is meaningless, but a Peugeot with 240,000 km is worth precisely €3,650. That’s what your model tells me. The absurd never stood a chance." — Camus, if he’d survived and learned Python

So if you ever looked at a rusty 200k-kilometer car and whispered, "How much are you worth, really?" — this one’s for you.

Before you embark on this beautiful, soul-crushing journey, clone the repository and install the dependencies:

git clone https://github.com/tiboitel/ft_linear_regression.git
cd ft_linear_regression
chmod +x install.sh
./install.sh
source .venv/bin/activate

Pro tip: Don't forget to activate your virtual environment. Otherwise your machine might file a lawsuit for dependency abuse.

Train the model:

# Default usage
python3 train.py

# Custom learning rate
python3 train.py --alpha 0.01

# Custom iterations
python3 train.py --iterations 5000

# Both
python3 train.py --alpha 0.1 --iterations 3000

Sit back, relax, and watch your machine learn what life has already taught you — more mileage = less money.

Make a prediction:

python3 predict.py

Type in a car’s mileage. Get an estimated price. Panic at how accurate it is.

Visualize the magic:

python3 plot.py

See the data. See the line. Understand your own fate.

In train.py, we make the model learn from the truth buried in a CSV of past car sales:

• Normalize the mileage so your model doesn’t get intimidated by big numbers.

• Start with a dumb guess (a flat line).

• Slowly adjust that line by figuring out how wrong the guesses are and fixing them — step by painful step.

• Save the learned formula into models/model.json.

No libraries. No mercy. Just raw numpy, pandas, and brain work.

In predict.py, you whisper your car’s mileage into the void, and the model speaks back — with a price.

• Behind the scenes:

• It grabs the trained values.

• Normalizes your input like a boss.

• Applies the sacred formula it learned from the ancients (aka your training run).

• Prints the predicted price with more confidence than your last relationship.

In plot.py, we:

• Plot your data points (every sad little car).

• Draw the regression line (the oracle of used car value).

• Make it look decent with labels and all that jazz.

Sit and admire the curve of capitalism.

├── data
│   └── data.csv        # The holy scripture of KM and price
├── install.sh          # Setup the temple of wisdom
├── models
│   └── model.json      # The saved wisdom of your AI Jacky-tuning Monk.
├── plot.py             # Visual art for your linear gradient
├── predict.py          # Ask your modern pythia
├── README.md           # You're here. Congrats !
├── requirements.txt    # Dependencies. Blame then when it break.
├── src
│   ├── config.py       # Assure to don't lose your PATH
│   ├── data_loader.py  # The scribe of the scripture
│   ├── __init__.py     # Could be useful one day
│   ├── linear_model.py # The temple of learning
│   ├── model_io.py     # How the pythia read the stars
│   ├── plotter.py      # The Artist
│   ├── predictor.py    # The Magician
│   └── utils.py        # Whatever you can't put elsewhere
└── train.py            # Access door of the temple

Ashen one, when it comes time to run the tests, don't sweat it. It's relatively simple. Just run the command:

pytests test/

Enjoy the little green fields and the beauty of well-written code while sipping your Estus in front of your screen.

This project is proof that you don’t need TensorFlow to make machines think — just grit, sweat, and numpy (optional but, no one like to normalize data and write standard deviation).

So go ahead, predict some prices. Plot some lines. Break some hearts.

And remember:

"The mileage may vary, but the good car is the one you love." - Captain Falcon

Licensed under the Entropy of the Void. You can use, modify, and break this code however you want — just don’t blame me when it starts replying with existential questions.

Written by tiboitel, Python sorcerer initiate learning forbidden ML magic.

If you enjoyed this, wait to see the day I tackle neural networks to see a new era of softened brains ready to devour themselves.

See you Space Cowboy 🤠 🚀