Thank you to all who attended PyData London 2014. We had a great time meeting up with everyone, attending talks, and learning more about the cool things being done with Python and Data these days. We were pleased to see many of our own team participate – from tutorials and talks to networking.
Many of you have probably already heard about the PyData conferences that traditionally took place only in the US. And you probably already know that it’s one of the best places to exchange experiences about Python and Data. Well, last month, PyData hit Europe! On February 21-23, 2014 the first ever PyData London took place. Why London? Well, London is one of the main financial hubs in the world, and the UK is well- known as a country where many scientific applications are designed and implemented. So, it made a lot of sense for PyData conferences to be introduced to Europe here.
The conference itself was three days long, with tutorials on the first day and talks on the other two. Tutorials are a great way to give people an introduction to the use of Python for processing, analyzing, and visualizing Big Data (one of the main goals of the PyData conferences). The tutorial sessions started with an excellent tutorial by Yves Hilpisch, Managing Director of Continuum Analytics Europe, on “Interactive Financial Analytics with Python and IPython“, where people were introduced to the task of analyzing financial data in a step-by-step manner. All of the materials were delivered in the form of IPython notebooks, so that attendees were able to follow the explanations at their own pace. The tutorial was extremely well received, and many people were enthusiastic about the new doors that Python and IPython are opening.
The other tutorials ranged from introductory level, like “Recommenders in Python” to advanced, like “Embeddings in Python”. With a few intermediate level talks in between, including “Plotting Large Datasets for the Web with Bokeh” from Continuum’s own Bryan Van de Ven. In this tutorial, the new features of the Bokeh library were described in depth, showing data analysts new tools for interactive data visualization. The “Databases for Scientists” tutorial by David Cournapeau and Simon Jagoe was also interesting. Data scientists were introduced to the intricacies of relational databases in Python with a lot of good advice and common sense (you know, the least common of the senses :).
On Saturday, the talks started with a keynote by Felix Fernandez from Deutsche Börse. He gave a very entertaining and broad overview of how Python is taking many financial places by storm, notably Deutsche Börse in Frankfurt, Germany. Felix explained how Python is being used internally for many tasks, from systems administration to data processing and visualization, with excellent results.
After this, I talked about how to make programs more efficient memory-wise through a judicious use of compression, notably in scenarios where a lot of data needs to be loaded in-memory at the same time. I showed that, in the future, computing time could benefit from using compression too. Other presentations given on Saturday covered topics such as data visualization, machine learning, databases and high performance computing for Big Data.
The last day was kicked off by Gaël Varoquaux, researcher in brain imaging and modeling at INRIA in France. He presented a personal perspective on ten years of scientific data processing with Python and his opinion on the emerging patterns in data processing. Very inspiring.
During this day, Yves Hilpisch gave a talk on derivatives analytics with Python. He briefly reviewed some important pieces of option pricing history and then went on to show some simple and also a rather complex example, valuing a multi-risk, multi-derivatives portfolio by the use of DX Analytics. DX Analytics is a derivatives pricing library completely implemented in Python and using Monte Carlo simulation for pricing and hedging.
The Big Picture
All in all, there were many excellent talks and tutorials indeed, and the conference excelled in its main goal of gathering people who share the similar interests and allowing them to exchange their experiences and enable future collaborations. The organizers (thanks Ian Oszvald and Leah Silen for the excellent work on that front) did a great job creating a relaxed atmosphere of friendliness that aided in facilitating interaction between attendees. The conference venue provided by Level39 in the Canary Wharf area had breathtaking views of the Thames and other surrounding skyscrapers, which contributed to the feeling of being in a special place for a special event.
We hope that although PyData London 2014 is over, the attendees will continue to find it useful in improving their knowledge of Python in Big Data and for extending their network, ultimately making collaboration easier and more productive.
PyData Silicon Valley and Berlin
BTW, PyData just announced two upcoming conferences. PyData Silicon Valley on May 2-4 at the Facebook headquarters in Menlo Park, and PyData Berlin, in conjunction with EuroPython, on July 25-27. Hope to see you there!