Skip to content

Analysing Jupyter Notebook 2015 Survey Responses

TL;DR ⏩

From my analysis, I deduced that in 2015:

Jupyter notebook users would have loved tutorials and documentations to guide them through the app.

Jupyter Notebook users needed improved interactivity and integrations with ipython, git, github, matplotlib, etc.

Running the app in the browser made it difficult to use.

Jupyter Notebook did not help with debugging, version control, and editing.

Skip to 💨

About the Project

Jupyter Notebook is a flexible web app that allows you to compute data, write various programming languages, plain text, equations, markdown, etc. For my first data analysis project, I analysed and explored the responses from a 2015 Jupyter Notebook survey organised by Project Jupyter with python. You will find the dataset here.

🔗 You can view the more detailed analysis (with code) here

Findings: Here’s what I found…

The survey had 1706 responses and 34 questions. Only 779 (46%) responders completed the survey. I only worked with the complete responses.

Completion vs. Partial completion of the survey

Most participants were:

  • Data Scientists
  • Students
  • Researchers
What is your primary role when using Jupyter Notebook (eg., student, astrophysicist, financial modeler, business manager, etc.)?

They belong to the Research, Education, Finance and Science industries.

49% of participants used Jupyter Notebook Daily.

How often do you use Jupyter Notebook?

75% had been using it for more than a year.

Roughly how long have you been using Jupyter Notebook?

87% ran Jupyter Notebook as a standalone app.

How do you run Jupyter Notebook?

Among the 8% of participants that chose to write their response, most used docker and SageMath to run the app.

Other – Write in: How do you run Jupyter Notebook?

59% of participants have less than 10 people viewing their notebooks.

How many people typically see and/or interact with the results of your work in Jupyter Notebook? (Consider people who view your notebooks on nbviewer, colleagues who rerun your notebooks, developers who star your notebook repos on GitHub, audiences who see your notebooks as slideshows, etc.)

The Good

For some, Jupyter Notebook, at the time, was easy to use. The markdown, integration and inline features added to its convenience.

Aspect 1-3:What aspects of Jupyter Notebook make it pleasant to use in your workflow?

Participants found it useful for its interactivity, data analysis and visualisation, markdown, and integrations.

Workflow Need 1-3: What needs in your workflow does Jupyter Notebook address?

Despite its integration and interactivity strength, most participants mentioned that they would like git, github, matplotlib, Vim, Emacs, etc to be tightly integrated with the notebook.

Tool / Application 1-3:What tools and applications, if any, would you like to see more tightly integrated with Jupyter Notebook?

Its integration with python made it pleasant to use.

72% of participants thought Jupyter Notebook was “convenient”.

Select all the words that best describe Jupyter notebook.

The Bad

For some responders, Jupyter Notebook was difficult to use. A few nice to haves were:

  • Tutorials
  • Documentation
Enhancement 1-3: Thinking back to when you first started using Jupyter Notebook, what enhancements would have made your initial experience better?

Poor version control & debugging. These hindered the workflow of Jupyter Notebook users.

Workflow Need 1-3: What needs in your workflow does Jupyter Notebook not address?

In addition to its editing capabilities, using Jupyter Notebook in the browser made it difficult to use.

Aspect 1-3: What aspects of Jupyter Notebook make it difficult to use in your workflow?

It’s a wrap…for now

This is not perfect. It isn’t even remotely close to perfect, but the process was worthwhile. I will appreciate your feedback.