machine-learning-helpers / induction-python

Python source code derived from books about Machine Learning (ML)

Overview

This repository features a set of various Python Jupyter notebooks, collected from books, blogs, or originally designed and developed by the organization team. It provides guidance for end-to-end working examples of how to author Python Jupyter notebooks in just a few minutes.

The source code is properly attributed in the corresponding directories. If you believe some attribution is missing, please submit a pull request or an issue.

That project makes use of Jupyter Lab and Python virtual environments, which can either be:

• Installed locally on your laptop/workstation. More details are available in the corresponding sections of this project:

  • Python virtual environments
  • Jupyter Lab

• Run from/within Docker. More details are available on the Docker images for Python Jupyter Lab notebooks project

• More integration with Cookiecutter Data Science may happen in the future. Those guidelines are nevertheless worth the read. As of February 2019, Cookiecutter Data Science works with explicit virtualenv and requirements.txt files, whereas we believe that Pyenv and pipenv are now the right choices to manage Python dependencies (as pipenv also uses virtualenv, but in a more seamless way for the developer).

See also

• NumPy
• SciKit-Learn
• Pandas
  • GeoPandas
• Jupyter Lab
• Optimus, Agile Data Science Workflows made easy
• Featuretools
• Dask:
• Matplotlib
• Seaborn:
• Altair for visualization

Dependencies

That projects makes use of Jupyter Lab and Python virtual environments. More details are available in the corresponding sections:

• Python virtual environments
• Jupyter Lab

Cookiecutter Data Science

• A dependency on Cookiecutter Data Science has been added to pipenv. Starting a new project is now as easy as issuing the following command:

$ mkdir -p ~/dev/ml
$ pipenv run cookiecutter https://github.com/drivendata/cookiecutter-data-science
$ cp .python-version Pipfile <resulting-project-directory-structure>/
$ mv <resulting-project-directory-structure> ~/dev/ml
$ pushd ~/dev/ml/<resulting-project-directory-structure>
$ pipenv install && pipenv install --dev
$ pipenv run python ./test_environment.py
Loading .env environment variables…
>>> Development environment passes all tests!
$ git init .
$ git remote add origin https://<git-server>:/<your-preferred-repo>
$ git add .gitignore .python-version LICENSE Makefile Pipfile* README.md docs models notebooks references reports requirements.txt setup.py src test_environment.py tox.ini
$ git commit -m "Project creation"
$ git push --all
$ popd

• An example is provided in the ml_induction_python project folder