Jupyter Newsletter 6 • August 31, 2016
Altair, JupyterHub Workshop, Recent Events
Altair
One of the core Jupyter developers, Brian Granger, has recently started a new visualization project, Altair, with Jake Vanderplas. Altair, offers a declarative API for fast and efficient statistical data exploration and visualization. Announced at SciPy 2016, Altair’s aim isn’t to be another plot rendering tool among Python’s current visualization milieu, but rather an elegant and straightforward option for Python users to produce incredible and compelling visualizations without needing a significant amount of code.
Altair generates JSON-formatted visualization specifications following the well-defined Vega-Lite visualization grammar. Jupyter notebooks can then render this JSON data using the Vega-Lite, Vega and D3 JavaScript packages. By using this declarative, grammar-based approach, Altair could someday serve as a means for various visualization libraries to work together, communicating via the lingua franca of the Vega-Lite grammar.
Altair’s approach enables users to perform exploratory visualizations with a much simpler API initially, pick an appropriate renderer for their usage case, and then leverage the full capabilities of that renderer for more advanced plot customization. With Altair, you can spend more time understanding your data and its meaning, as well as creating the best methods of displaying that data for others.
JupyterHub Mini Workshop
Have you always been curious about supercomputers? Did you know they can be used through JupyterHub? This was one of many interesting topics covered by ten different speakers during the online JupyterHub workshop on June 22nd, 2016. The workshop was organized by F. Perez and J. Whitacre at UC Berkeley, along with C. Willing, A. Ruvalcaba and R. Neetro at CalPoly.
JupyterHub is a multiuser version of the Jupyter Notebook that is being used in a wide range organizations. This workshop covered the JupyterHub architecture, its deployment using a range of different approaches (ansible, docker, etc.) along with a number of different usage examples in education, research and commercial industry.
Want to learn more? To read and view workshop content please visit the GitHub repository for the workshop. We also have two videos of the talks on YouTube:
Part 1: https://www.youtube.com/watch?v=VnMSjR5IoS8
Part 2: https://www.youtube.com/watch?v=6p1xhi0P5dg
Recent Events
JupyterDay (Atlanta - August 13, 2016)
Our latest community led event was was recently held in Atlanta. The day kicked off with a thought provoking panel discussion, “Open Source is the Medium of Innovation,” led by O’Reilly’s Paco Nathan. Talks about tracking weather and storms, teaching math on YouTube, and Peter Norvig on O’Reilly’s Oriole project rounded out the morning. The afternoon shifted to talks on JupyterHub and JupyterLab and community workshops where people shared resources, projects, and ideas.
We wish to thank the generous sponsors (O’Reilly Media, Continuum Analytics, Georgia Tech Research Institute, General Assembly), attendees, and speakers. Event organizers, including Tony Fast, Nick Bollweg, and Ana Ruvalcaba, shared resources to create a great day of learning and collaboration for all attendees.
PyData SF (August 12 - August 14)
The NumFOCUS organized PyData Conference series returned to the bay area this August for PyData SF at the Galvanize office in SoMa. Over 200 attendees listened to Jupyter Technical Project Manager, Jamie Whitacre, give a talk about the next generation Jupyter user interface, JupyterLab. Brian Granger gave a talk about Altair (see above), and Stuart Geiger gave a talk about his work studying the Jupyter and IPython GitHub organizations using data from GitHub.
PyBay (August 19 - August 21, 2016)
Several Project Jupyter contributors attended the inaugural PyBay Conference in San Francisco, California. Cal Poly San Luis Obispo student and Jupyter contributor, Roshan Prabhakar, supported the organizing conference committee by volunteering his time. Jupyter Technical Project Manager, Jamie Whitacre, presented an overview of the entire project and development activities in progress including iPython 5.0, JupyterHub, and JupyterLab. Core Jupyter developer, Matthias Bussonnier, gave a notebook demo and spoke about Xonsh.
EuroSciPy (August 19 - August 21, 2016)
At this year’s EuroSciPy, several talks focused on Jupyter. Below you will find short summaries of the topics presented:
“Jupyter and its Horizons”
Ever since Jupyter split from IPython in 2015, it has grown into an active collection of projects. Min Ragan-Kelley gave a talk about what’s new and coming in Jupyter and the wider community, including JupyterLab, kernel gateway, dashboards and other incubator projects, in addition to community projects such as nteract and rodeo.
“Can we publish Jupyter Notebooks as academic papers?”
Scientists in a varying number of fields have published papers with accompanying notebooks, allowing readers to reproduce their data analysis. However, no-one has yet published a notebook as a peer reviewed paper. Thomas Kluyver gave a talk detailing the role of notebooks in academic publication, including impediments to its use and possible solutions to challenges such as citations.
“Nbdime - Notebook diffing and merging”
Vidar Tonaas Fauske discussed nbdime, a new project to help with diffing and merging notebooks, including integration with version control systems such as git.
Featured Contributors
Jamie Whitacre, Technical Project Manager, works with the core Jupyter team on organizational structure, software process/management and in strengthening connections with the larger Jupyter community. She is located at the Berkeley Institute for Data Science (BIDS) and is an affiliate of the Lawrence Berkeley National Laboratory. Before joining the project, Jamie worked with the Smithsonian’s National Museum of Natural History’s Informatics team designing and building scientific data systems for the museum’s genomics and biodiversity research initiatives. She has also worked in industry including health IT and biotechnology.
Spoorthy Vemula is a User Interface/User Experience Designer and Software Engineer for Jupyter at Cal Poly. He is a Software Engineering major and is going to be a sophomore at Cal Poly this fall. Before working at Jupyter, Spoorthy was an iOS developer and designer and made the app “5minShower”. At Jupyter, he is working on team IO to help redesign the interface of JupyterLab. You can find him on GitHub as @spoorthyvemula and you can see the rest of his work at www.spoorthyv.com.
Upcoming Events
PyData Carolinas (September 14–16, 2016)
PyData is brought to you by NumFocus and is a Python conference that brings together both users and developers of data analysis tools to share ideas and learn from each other.
PyCon UK (September 15–19, 2016)
PyCon is hosted by the Python Software Foundation and is one of the largest annual gathering of the developers and users of Python. More details are on the conference website, where you can buy a ticket or submit a proposal for a talk or a workshop.
Latest Developers Meeting: August 30, 2016
A special thank you to all who collaborated to create this newsletter: Brian Granger, Carol Willing, Charnpreet Singh, Farica Carroll, Fernando Perez, Jamie Whitacre, Matt Bowers, Reese Netro, Roshan Prabhakar, and Spoorthy Vemula.
