Why open-source software matters
Computer software can often cost more than the machine itself. Not only is there a huge economic barrier for anyone looking to explore a new tool, program, or application, but anyone interested in taking bits and pieces of a software’s offering is deterred by exorbitant price points. Furthermore, if something ever goes wrong with an installed application, little to no support is available, other than upgrading to the next costly version. It’s no wonder that in recent years, open-sourced software has become increasingly important in the world of data science, and inter-disciplinary computer science.
Open sourced software allows for greater transparency software creation and maintenance. Users can open up a piece of software and “look under the hood” to study software’s inner workings. Users can also modify an application, allowing customization not available with traditional software. But the greatest draw of open-sourced software is its price point. With most programming languages entrenched in costly software options, Python has become the primary language of open-source solutions in data science. Most python download’s are free or low-cost, and if you have access to a computer, even through a public library, you can start programming with Python. What gives Python its life and sustainability is a community of people constantly willing to improve and maintain python-based software, and applications. Enter: SciPy.
The Wonderful World of Sci-Py
SciPy is an ecosystem using a collection of open-source software for computing in Python, and an international community continuing to improve upon, and develop this software. What started off as a group of self-described “odd-balls” has grown into a fully bonafide annual conference with 600 attendees from all over the United States, and representatives from 15 countries. Speakers from previous SciPy conferences have included Python superstars such as Guido van Rossum, and the developers of the iPython and NumPy packages that are a part of the SciPy stack.
This Year, NYU’s Center for Data Science attended and presented at the SciPy conference in Austin, TX, along with the two other Universities comprising the Moore-Sloan foundation.
CDS Takes Austin
Even though it was their first year, the three schools in Moore-Sloan partnered with SciPy for an unparalleled amount of representation. A widely recognized name in the world of machine learning, NYU’s Andreas Mueller helped kick off the conference with an extensive tutorial on machine learning. Andreas focused on Scikit-learn, an open-sourced python library where he serves as the lead developer. Later Andreas gave a conference talk on structured prediction in PyStruct, another python based library he helped develop. The Center for Data Science’s other fellow in attendance, Brian McFee, presented his work on Librosa, an open-source audio and music signal library he developed for his own research in the field of music information retrieval.
Overall, the common thread in SciPy was a furthering of the open-sourced nature of the Python community. Academia thrives on work and research that is by nature open-sourced, free, and accessible, and the SciPy community in conjunction with organizations like Moore-Sloan continue to promote these ideas in data science and beyond.
By Jack Lowery