Repo for the Applied DS Book

This is currently a work in progress, but feel free to take a look: https://deil87.github.io/applied_data_science_book/intro.html

Local installation

source adsb_env/bin/activate  
pip install -r requirements.txt 

In order to build the book run this in the root directory:

jupyter-book build applied_data_science_book/.

(known issue, LTR folder is automatilly being picked up and we receive warnings)

Remote deployment to Github Pages

Deployment is done through Github Actions. See /.github/workflows/static.yml

Getting started

Adding a chapter or section:

See _toc.yml

Jupyter-book will find corresponding .ipynb or .md files and render them as part of the book

Dependency hell

jupyter-book might have many python dependencies as for each page/jupyter notebook we might want to have different dependencies, is it correct?

Yes, your understanding is correct: a Jupyter Book might require many Python dependencies because each page or Jupyter notebook could potentially need different libraries or versions.

Why is this the case?

• Each notebook in a Jupyter Book can contain code that depends on different Python packages depending on the analysis or subject.
• Unlike a single script or project, you may have notebooks covering diverse topics, like data visualization, machine learning, or bioinformatics, each requiring specific libraries.
• This can lead to a larger, more complex set of dependencies overall to ensure every notebook runs properly.

Common approaches to manage this:

• Single consolidated environment: You collect all dependencies needed across all notebooks in one big environment (e.g., one requirements.txt or Conda env) so that any notebook can run without missing packages.
• Multiple environments: You may maintain separate environments for different parts of the book if notebooks are very different (more complex to maintain).
• Use of Binder or Docker: For reproducibility, Jupyter Book supports integration with tools like Binder or Docker where you specify dependencies in an environment file; this installs everything upfront.

Practical advice:

• It's best to list all required packages explicitly in a main environment file.
• For maintainability and reproducibility, pin exact versions of dependencies.
• Consider tools like pip-tools or Conda environment files to simplify managing complex dependencies across notebooks.

---

Summary:

• Jupyter Book can have many dependencies collectively, as each notebook may require different Python packages.
• Often all are installed in one environment for simplicity.