ML Project Generator creates a ready-to-use ML project skeleton with directories for data, models, and notebooks, basic Python files, a .gitignore, and a list of dependencies. Start with a general-purpose template or choose a structure designed for Scikit-learn or PyTorch.

Features

• three templates: basic, sklearn, and pytorch;
• automatic creation of directories and files;
• automatic requirements.txt generation;
• optional Jupyter dependency with --jupyter;
• optional Matplotlib and Seaborn dependencies with --plotting;
• dedicated README and .gitignore files for generated projects.

Requirements

You need:

• Python 3.10 or newer;
• pip;
• Git, only when installing by cloning the repository.

Check your Python version:

python --version

Installation

Quick installation on Windows

Clone the repository and run the installer:

git clone https://github.com/KishlakEnjoyer/ml-project-generator.git
cd ml-project-generator
.\install.bat

The installer will automatically:

• verify that Python 3.10+ is available;
• install pipx;
• add the CLI application directory to PATH;
• install the ml-init command.

Close and reopen your terminal after installation. You can then run ml-init from any directory:

ml-init --help

Run install.bat again to update the installed command from your current local copy of the repository.

Manual installation for development

An editable installation is more convenient when changing the source code. Create a virtual environment:

python -m venv .venv

Activate it.

Windows PowerShell:

.venv\Scripts\Activate.ps1

Linux and macOS:

source .venv/bin/activate

Install the tool:

python -m pip install -e .

The ml-init command is now available in the active environment.

Quick start

Create a basic ML project:

ml-init my-project

Create a Scikit-learn project:

ml-init churn-prediction sklearn

Create a PyTorch project with Jupyter and plotting libraries:

ml-init image-classifier pytorch --jupyter --plotting

The project is created in the current directory. For example, the last command creates an image-classifier directory.

Usage

ml-init NAME [TEMPLATE] [OPTIONS]

Arguments

Argument	Description	Default
NAME	Name and directory of the new project	required
TEMPLATE	Template: basic, sklearn, or pytorch	basic

Options

Option	Short form	Description
--jupyter	-j	Add jupyter to the project dependencies
--plotting	-p	Add matplotlib and seaborn
--version	-v	Display the tool version
--help		Display command help

Display the built-in help:

ml-init --help

Templates

basic

A minimal, general-purpose structure for experiments and small ML projects. It contains data, models, notebooks, src, and tests directories.

ml-init experiment basic

sklearn

A structure for classical machine learning projects. It includes separate files for data loading, preprocessing, training, evaluation, and prediction.

ml-init tabular-model sklearn

Main dependencies: NumPy, Pandas, Scikit-learn, and Joblib.

pytorch

A structure for deep learning projects. It includes files for the Dataset, model, training, evaluation, and inference, along with checkpoints and outputs directories.

ml-init neural-network pytorch

Main dependencies: PyTorch, NumPy, and tqdm.

After generating a project

Enter the generated directory:

cd my-project

Create a separate virtual environment:

python -m venv .venv

Activate it on Windows PowerShell:

.venv\Scripts\Activate.ps1

Or on Linux and macOS:

source .venv/bin/activate

Install the generated project dependencies:

python -m pip install -r requirements.txt

You can now place source datasets in data/raw, experiments in notebooks, and reusable application code in src.

Example generated structure

my-project/
├── data/
│   ├── raw/
│   └── processed/
├── models/ or checkpoints/
├── notebooks/
├── src/
├── tests/
├── .gitignore
├── README.md
└── requirements.txt

The exact set of files depends on the selected template.

Development

Install the project in editable mode together with Pytest:

python -m pip install -e . pytest

Run the test suite:

pytest

Preset files are located in src/mlproject/presets. Each template defines three values:

• DEPENDENCIES — dependencies for the generated project;
• DIRECTORIES — directories to create;
• FILES — files to create and their contents.

License

This project is distributed under the MIT License.