Our October meetup is hosted by Babbel in their new offices in Mitte. The meetup is going to have a slightly different structure than the usual meetups.
Beginning of October we organised a hackathon at Data Science Retreat around the theme of predicting political views from text. The event was a follow up to Felix Biessmann's talk at the PyData conference in May (https://www.youtube.com/watch?v=IhUSiXXg4rg). After two days of hacking away we ended up with lots of ideas and several started projects. This meetup is going to feature presentations by those who took part about what they did and how to carry to work forward. We'll aim to keep each talk reasonably short, about 10 to 15 minutes max.
19:00 : Doors open, food and drinks provided by Babbel
19:30 - 21:30 : Presentations and discussion
- Matti Lyra / Felix Biessmann: Introduction
- Matti Lyra: Dask and neural networks for classification
I'll talk a little bit about `dask` and its pretty cool data loading capabilities, and then move on to my rather unsuccessful fumbling with `keras` and implementing neural networks (because deep learning!)
- Katherine Jarmul: How not to DDoS the European Parliament
- Elad Verbin: How the words you use reveal your political affiliation
- Jirka Lewandowski: Contextual word usage of political parties
- David Mariassy: Topic-labelled political speeches
- David Batista: Issue focussed sentiment analysis of political parties
- Felix Biessmann: Building an active learning tool to gather better annotations
- Alberto Granzotto: Facebook.com: Popping the Filter Bubble