Notebooks-to-Packages

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

Notebooks-to-Packages

(★ 16)

course material for the "Notebooks to Scripts to Packages" workshop as part of Princeton Wintersession 2023

Notebooks-to-Packages Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Turning your Python notebooks into Python packages #

In this workshop we will take a data analysis pipeline implemented in a Jupyter notebook and convert it into a script
that can be run from command-line. We will then convert this script into a Python package: a collection of code modules
supporting a pre-defined set of command-line tools. Finally, we will refactor the package by applying the paradigm of
object-oriented programming.

The purpose of this course is **not** to dissuade you from using Jupyter! Notebooks are an incredibly accessible and
powerful tool for data scientists and researchers alike. However, as an experiment expands in scope and scale, the
limiting features of notebooks start to become readily apparent. We will focus on the _process_ of software design:
where and when in the course of building an analysis pipeline you may want to consider investing the effort to leverage
the other tools at your disposal as a Python developer.

## Table of contents ##

1. **Notebooks**
    - predicting UFO sightings, as implemented in a Jupyter notebook
    - the advantages and disadvantages of notebooks
    - when in the development of an experiment to consider moving beyond a notebook

2. **Scripts**
    - converting a notebook into a script  
    - parametrizing a script using `argparse`
    - modularizing a script using helper functions

3. **Packages**
    - how packages are designed in Python
    - possible ways to structure your package
    - creating package infrastructure
    - sharing your package with the world

4. **Classes**
    - applying object-oriented programming within a package
    - how OOP affects package structure
    - refactoring a class design to introduce hierarchical class structure

## Preparing for the workshop ##

These materials are designed for users with at least some knowledge of Python, and particuarly with using Jupyter
notebooks to build data analysis experiments. You may also want to refresh your acquiantance with the use of Python
packages such as `requests`, `pandas`, `matplotlib`, and `scikit-learn` before starting this workshop.

To run the code included in this workshop, you'll need access to a command-line environment with a
[conda](https://conda.io/projects/conda/en/latest/index.html) installation. In this environment, choose a place to check
out the course repository:

```git clone git@github.com:michal-g/Notebooks-to-Packages.git```

In the newly-created folder `Notebooks-to-Packages` you'll find the workshop materials including the code; we create the
the environment to run the code and activate it using:

```
conda create --name notebooks-packages -c conda-forge python=3.9 pandas plotly jupyter imageio matplotlib \
                                                                 'scikit-learn<1.1' nbconvert nbformat
conda activate notebooks-packages
pip install kaleido skits
```

## Past versions ##

- `v1` presented as part of Princeton Wintersession 2023

## Notes ##

The dataset `nuforc_events_complete.csv` was downloaded from Link Wentz' [repo](github.com/LinkWentz/NUFORC-Dataset) on
January 12, 2024.

Notebooks-to-Packages

Notebooks-to-Packages Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Notebooks-to-Packages

Notebooks-to-Packages Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README