Wrap-up, Outlook, Jupyter and Quarto

Statistical Programming
Prof. Dr. Martin Spindler

Table of contents

• Notebooks & Publication Systems
• Literature

It’s a wrap!

Congratulations

🥳 You’ve learned how to program with Python! 🎉

What’s next?

• I’d encourage you to practice programming and develop your programming skills further!

• Here are some resources

  • freeCodeCamp - Scientific Computing with Python
  • Tiny Python Programs
  • Python for Data Analysis
  • kaggle competitions for beginners
  • Advent of Code

• Start your own project

  • Try to use Python in your work / thesis!

What’s next?

Statistics, Causal ML

• Huber (2023)

• Cunningham (2021)

• Chernozhukov et al. (2024)

• Facure (2023)

What’s next?

• Learn more tools / skills
  • Git / GitHub
  • Data Products with Quarto
  • More of the Python Ecosystem

Notebooks & Publication Systems

Notebooks and Publication Systems

• Jupyter and Quarto are two frameworks to combine data and code in a single document

  • Jupyter is a notebook framework that is based on Python
  • Quarto is a framework that is based on markdown and can be used with Python, R, julia, etc.

• Both frameworks are based on markdown and allow to combine code and data in a single document

• You can use them to create reports, blogposts, websites, slideshows, etc.

• You can load data, run code, and display (& polish) results in the same document

Jupyter

• Installation guidelines & getting started guide
  • https://jupyter.org/install
  • https://jupyter.org/try
• Open Jupyter Notebook and create a new notebook (.ipynb file)
  • Open Anaconda Navigator and launch Jupyter Notebook
  • Create a new notebook
  • Run some code (e.g., from lecture or problem set)
• Alternatively, you can use VS Studio Code
  • Open Anaconda Navigator and create a new .ipynb file
  • Create a new notebook
  • Run some code (e.g., from lecture or problem set)

Jupyter

• A Jupyter notebook has two types of cells:
  • code cells, and
  • markdown cells
• Markdown cells allow you to write text and format it using markdown syntax
  • Basics of markdown in Jupyter
• Code cells allow you to write and run code
  • You have to specify the a kernel for the programming language in the top right corner

Jupyter

• Jupyter is very popular and it easily integrates with cloud services/servers

• Very convient: google Colab!

  • You can run notebooks in the cloud (e.g., on Google servers)
  • You can share notebooks with others
  • You can run notebooks on GPUs (e.g., for deep learning)

• There are various extensions available, see https://docs.jupyter.org/en/latest/

  • For example nbconvert to convert notebooks to other formats (e.g., html, pdf, etc.)

Quarto

• Quarto is a relatively new framework to generate dynamic content, like

  • (Automated) Reports that involve Python, R, julia or other code
  • Blogposts, websites, pdf files, slideshow, jupyter notebooks, interactive apps, MS office documents
  • Documents including references (e.g. .bib files) and math formula (LaTeX)

• Thereby, Quarto has a unified syntax that is based on markdown

• Installation guidelines & getting started guide

  • https://quarto.org/docs/get-started/
  • For Python: VS Studio recommended
  • Open VS Studio and install Quarto extension

Quarto

• Open new .qmd file and render
  • Render using the VS Studio features (Ctrl + k) or command line quarto render
• Go through the Quarto Guide to find out more
  • Topics: Basic syntax, computations, document types, output type (tables, figures, etc.)

Quarto

• It is very easy to get started with Quarto

• You can use Quarto in combination with GitHub, for example to set up your personal website

• Quarto is developing very fast, for example recently they added support for dashboards

Quarto: PDF files

• Create a new .qmd file and render it as .pdf
  • Header option format: pdf
  • You can source a .bib file and use LaTeX syntax to cite references
  • You can also use a .tex preamble and commands to customize the document
  • You can use the quarto render command in the terminal
  • You can also use the VS Studio features (Ctrl + k) to render the file
• You can also output .docx files etc.

Quarto: Presentations

• You can use Quarto to create slideshows, see https://quarto.org/docs/presentations/
  • .ppt
  • .html (through reveal.js)
  • .pdf (through latex beamer)

• With reveal.js, slideshows become interactive, for example, you can include GIFs, videos, animations, etc.

• reveal.js: Create a new .qmd file and render it as .html slideshow

  • Header option format: revealjs

via GIPHY

Literature

References

• Jupyter website https://jupyter.org/

• Quarto website https://quarto.org/

References

Chernozhukov, Victor, Christian Hansen, Nathan Kallus, Martin Spindler, and Vasilis Syrgkanis. 2024. “Applied Causal Inference Powered by ML and AI.” https://arxiv.org/abs/2403.02467.

Cunningham, Scott. 2021. Causal Inference: The Mixtape. Yale university press.

Facure, Matheus. 2023. Causal Inference in Python. O’Reilly Media, Inc.

Huber, Martin. 2023. Causal Analysis: Impact Evaluation and Causal Machine Learning with Applications in r. MIT Press.