At JupyterCon in New York last week, DataScience.com CSO William Merchan delivered a keynote address about the growing enterprise adoption of Jupyter, an open source web application that allows users to create interactive notebooks containing code, data visualizations, text, and more. You can watch the keynote in full below.
Usage of this powerful tool, which can be integrated seamlessly with the DataScience.com Platform, has increased ten times over from 2013 to 2017. In fact, in November 2016, Jupyter was downloaded 122,000 times; in July 2017, that number was over 300,000, according to data from the Python Package Index.
"It's exciting to see the community's growth, the community's involvement, and the continued innovation around the Jupyter ecosystem," Merchan said. "It's becoming the tool of choice. We're seeing this with the customers that we work with."
Merchan described three possible reasons for Jupyter's rise, which has contributed to a more cohesive and efficient data science process at many companies. The reasons he pointed to were:
1. Data science is now a critical business function: If done correctly, data science work now directly impacts every department of a business. No longer are data professionals relegated to a dusty corner of the office; now their work informs marketing, financial, and executive decisions. Tools like Jupyter notebooks, which are shareable and interactive, make sense in this new context.
2. The data community is rallying around Jupyter: Rather than requiring data scientists to learn proprietary tools, companies are increasingly creating data science frameworks that incorporate a variety of existing open source options. Jupyter is well loved and the number of new Jupyter packages being created makes that clear.
3. Python is king — and Jupyter was born out of Python: More than half of data scientists today use Python, according to research from conference organizer O'Reilly Media. Because Jupyter was born out of the IPython project, which created a powerful architecture for computing in the popular language, it's ideal for a large swath of data science work.
Merchan described the adoption of tools like Jupyter as part of a bigger movement toward alignment between data scientists and engineers, scalable processes, and, ultimately, data science platform adoption.
"If you grow from a team of one or two data scientists to hundreds or thousands, not having a consistent, cohesive technology approach to how you interact with all of these systems leads to trouble down the road," Merchan explained.
"[Platforms] need to connect into all of the tools data science teams prefer to use," he added. "...And they need to do it in a consistent way, in a repeatable way, that's in line with the technology and engineering teams' requirements as well."
Want to learn how to launch a Jupyter session equipped with deep learning tools in the DataScience.com Platform? Check out this video demo.