---
title: Why Use Jupyter Notebooks?
author: Colton Grainger
---

I recently participated in a webinar (July 2017) hosted by Science Gateways
Community Institute (which exists to facilitate community sharing of
experiences, technologies, and practices at little or no cost!) These are my
notes and reactions.

## Background

Dave Rosoff introduced me to SageMathCloud in 2014 for solving systems of
differential equations, which was my first exposure to IPython notebooks. As I
didn't know TeX then, I struggled tremendously to format my output. I didn't
pry deep into python's libraries; I mostly stayed in the sage docs. 

I was tasked to analyze resonance given different forcing functions for 2nd
order linear DEs. At finish, I had a mish-mashed assemblage of sage commands
and non-compiling markdown. I couldn't even plot a collection of functions
without manually typing in new the parameters (as opposed to now, where I'd
just plot a list of functions indexed by parameter values).

```python
plot([exp(-3*x)*(x^2 + C) for C in [-1..2]], (-0.5, 0.5))
```

## Webinar

Carol Willing presented a whirlwind tour of Jupyter's development out of IPython:

- For information about IPython prior to 2011, 
  see [Perez's historical ruminations](http://blog.fperez.org/2012/01/ipython-notebook-historical.html)

- In 2013, the IPython project started engaging users of 
  [Music21](http://web.mit.edu/music21/doc/about/what.html) 
  with interactive musicology notebooks. 

- In 2014, to break down barriers to entry, the IPython project rebranded
  itself as Project Jupyter (where IPython continued to provide the kernel), 

    - Andrea Zonca's [blog](zonca.github.io) was instrumental in promoting the
      project. (He continues to write about high performance computing and to
      offer Python tips.) 

- In 2016, the original author of IPython, Fernando Perez, and contributor to
  the notebook interface, Brian Granger, won substantial funding for
  [development](http://vcresearch.berkeley.edu/news/project-jupyter-gets-6m-expand-collaborative-data-science-software):
  "to serve users and foster collaboration". 

    - Perez cites Richard Hamming's 1962 observation that "the purpose of
      computing is insight, not numbers" as historic impetus for IPython
      notebook and motivation for the current form of the Jupyter notebook
      ([video](https://youtu.be/4pnj-eNEaqk)).

Following, Carol mentioned that Project Jupyter is determined to reach wider
audience (not just mathematicians and computer scientists). This bears
similarity to SageMathCloud's [rebranding](https://cocalc.com/). To increase
cross-disciplinary collaboration, Carol emphasized the notebook interface's
strengths in *data exploration, visualization, and analysis*.

### What's new with Jupyter?

- One could look at [trending notebooks](https://github.com/trending/juptyer-notebook).

- I can compile and share my notebooks with [Binder](https://beta.mybinder.org/).

    - to build github repos into live notebooks with the python kernel
    - won't work for notebooks running on a sagemath kernel

- There's [JupyterHub](https://jupyterhub.readthedocs.io/en/latest/) for online collaboration.

    - info sharing, passing off notebooks, browser based
    - also implemented [here](https://github.com/harvard/cloudJHub) on Amazon Cloud for teaching
    - to get started [Zero to JupyterHub](http://zero-to-jupyterhub.readthedocs.io/en/latest/)

- In design: [Jupyter Lab](https://github.com/jupyterlab/jupyterlab), and IDE.

- diffing and merging Jupyter notebooks with [nbdime](https://github.com/jupyter/nbdime)

- assigning and grading notebooks with [nbgrader](https://github.com/jupyter/nbgrader)

I am interested in Demba Ba's [Labs in the Wild](https://www.youtube.com/watch?v=h7WThtzuq_8) 
approach.  Demba emphasizes that data manipulation is part of literacy.

> [In a notebook] we expect to see data collection, experimental protocol,
> analysis, use of sophisticated algorithms, etc.

- He's taught a few vertically integrated signal processing courses with wearable    i
  [physiological monitors](https://www.empatica.com/e4-wristband) 
  and Jupyter notebooks.

- Example project: using wrist accelerator to predict the success of a freethrow.

    - Q: Can we set up an algorithm to provide feedback *during* the freethrow?