OTTO

Otto is a simple Boilerplate for Machine Learning projects integrated with MLflow that creates a basic directory structure to organize your code and data. Otto is strongly based on Cookiecutter, if you need something more complete is a good desition to visit they repo.

├── MLproject
├── README.md
├── Dockerfile
├── build_image.sh
├── .gitignore
├── data
│   ├── processed/
│   └── raw/
├── notebooks/
├── src/
│   ├── data/
│   │   └── __init__.py
│   │   └── data_preparation.py
│   ├── features/
│   │   └── __init__.py
│   ├── models/
│   │   └── __init__.py
│   │   └── model.py
│   ├── train.py
│   ├── settings.py
│   └── __init__.py
└── tests/

Usage

Simple install otto using pip as follows

pip install otto-ml

and use otto

otto --name new-project

or simple use it with out params and let otto guides you 😉

otto

and that's it, Now you can jump to code your model!

Ok, but... what this solve?

That is a pretty good question. The first attempt is to simplify the startup of a new machine learning project generating most, not machine-learning related code. Like the configuration of the docker image via Dockerfile or the MLProject setup and the connection with the Mlflow tracking server if you have set up one using ENV variables.

But to make it cristal water, let show how it will be a standard use of the otto package.

The Titanic Competition Example

... In development ...