21st Meetup

This will be a regular meetup (beer+pizza) at AHL like the previous months.

Note: Please use your full real names where signing up, otherwise we have problems with building security.

Our main speaker will be:

Rebecca Murphy (https://github.com/rebeccaroisin) and George Danezis (http://www0.cs.ucl.ac.uk/staff/G.Danezis/): Refreerank – Crowd-sourcing University Rankings with Scikit-Learn and NetworkX

Abstract: Last year, UK university departments took part in a review exercise (the Research Excellence Framework or REF) that aimed to evaluate the quality of their research. This process was hugely time consuming, expensive and somewhat subjective. We used scikit-learn and NetworkX to build a ranking system for UK Computer Science departments, based on publication data from the DBLP database. Our tool, Refreerank, was built in a weekend and reproduces a large fraction of the REF ranking, at no financial cost.

With Lightnings talks:

• Florian Rathgeber (https://kynan.github.io) (@frathgeber (https://twitter.com/frathgeber)): How to Git your (Jupyter) notebooks right

Abstract: Jupyter / IPython notebooks are great for having code and results in one place. But wouldn't it be great to be able to only keep the actual code cells in your Git repository without having to change your workflow? You're in luck! In this talk I will demonstrate nbstripout (https://github.com/kynan/nbstripout), a Python command line tool to effortlessly strip output from Jupyter notebooks, which doubles as a Git filter: you save your notebook as usual, including output, but Git never gets to see the output!

• Leonardo Baldassini (https://twitter.com/ImSoErgodic): Visualising intersections in Python

Abstract: The classical representation of Venn diagrams to visualise set intersections often becomes illegible when we need to intersect more than 3 or 4 sets. Lex et al. have recently introduced UpSet, a visualisation tool built in javascript that focusses on representing intersections as a bar chart coupled with an incidence matrix, allowing to overcome the limits of classic Venn diagrams (you can play with it here (http://vcg.github.io/upset/?dataset=0&duration=1000&orderBy=subsetSize&grouping=groupByIntersectionSize&selection=)). This talk will introduce PyUpSet (https://github.com/ImSoErgodic/py-upset), an open-source, static implementation of the core functionalities UpSet built on top of pandas and matplotlib.

• Daniel Slater (http://www.danielslater.net): Reinforcement learning in PyGame

Abstract: Google deepmind has recently had impressive results running reinforcement learning algorithms against Atari games. Here I present a frame work for running reinforcement learning on PyGame.

• Malcolm Sherrington (@JuliaLanguage, @JuliaBloggers): Quo Vadis, Julia?

The Julia project (http://julialang.org) is now almost 4 years old but still not yet at v1.0. This talk reviews the roadmap for increasing development in Julia, where it is at currently and whether it is ready for Enterprise work.

Logistics

Doors open at 6.30 (get there early as you have to sign-in via AHL's security), talks start at 7pm, beers from 9pm in the bar, We normally have > 200 folk in the room so there's plenty of people to discuss data science questions with!

Please remember to unRSVP if you realise you can't make it, we're limited by Security on the numbers so we need you to free up your place if you're not able to attend.

And make sure you follow @pydatalondon (https://twitter.com/pydatalondon) for any updates and early announcements. See you on the 1st at AHL!