PyData November Meetup

PyData Berlin
PyData Berlin
Public group

ResearchGate GmbH

Invalidenstr. 115, 10115 Berlin · Berlin

How to find us

Naturkundemuseum (U6), and Nordbahnhof (S1, S2 and S25). Please use the main entrance located on Invalidenstraße

Location image of event venue


The PyData Berlin November meetup is hosted at ResearchGate on Invalidenstraße. Doors open at 19:00, we'll start with a about half an hour of chatting and getting to know each other with drinks and snacks provided by ResearchGate.

After the event we'll regroup in Speisekombinat on Chausseestraße 116, 10115 Berlin at around 21:30 for more discussion.

As usual the event host is not a theoretical math equation and is alas restricted by finite space. Please therefore keep your RSVPs up-to-date if you can't make the event; be kind to your fellow PyData Berliners.

Speaker: Paul Boes

Abstract: Graphite is a service that lets you have more overview over the topics relevant to you, in less time. You feed it a bunch of texts that you are interested in and it creates a reading list from them such that following this list is the most efficient strategy for covering the contents of all the input texts. In my talk, I will introduce Graphite, its applications (currently mainly in the news sector) and discuss the details of the algorithm behind it. My main questions will be: What does "overview“ here mean? Is it reader-independent? How to best produce it? If I have enough time, I will also talk about our latest experiments with measuring the „novelty“ of articles.

Lightning Talks (5-10 mins):

Matti Lyra / PyData Berlin NumFocus, PyData and the numeric Python ecosystem

Sergio Alvarez Knipster is an online gaming platform for daily fantasy football. We're setting out to build a global fantasy football platform. The focus is to harness the massive amounts of real-time data generated during the football games to enable next generation fantasy football gameplays.

Adrin Jalali Attributing website traffic to TV-ads is tricky since we cannot track the source of those sessions back to the ad. I'll give a quick intuition about Gaussian Processes and show how to use them with `GPy` package developed at Sheffield's university to estimate the number of sessions coming as a result of a TV-ad.