Ninth PyData Lisbon Meetup

Hey everybody!

The sun is out and so are we. We’re happy to announce the ninth PyData meetup with 2 amazing speakers to tease your mind. Very exciting!

For this edition we’re having Bruno Dias from James for a second time (why change a winning team, am I right?). He'll be lifting the veil on discrete Bayesian networks for us.
As for the second speaker, we’re welcoming Vadym Safronov who will be addressing graph neural networks for multirelational data.
Is your interest peaked? Of course it is! You can find a detailed program below.

A huge shout-out to Zalando Tech for hosting the PyData meetup and providing us with drinks & snacks!

PROGRAM:

18.40 - 18.55: Walk-in

18.55 - 19.00: Greetings & intro

19.00 - 19.40: “Balls, Lines and Glorified Counts: Using Discrete Bayesian Networks to Synthesize Tabular Data” by Bruno Dias

With the advancements in Generative Adversarial Networks, we can now create simulations that produce high fidelity simulations and unseen/fake data. But, usually, this type of models is used with non-tabular data, lacking tools on how to deal with ordinal/nominal data and they also are “black boxes”, which makes it harder to adapt the data generation procedure. An alternative is to use Bayesian Networks. In this talk, I will introduce you to what a (discrete) Bayesian network is, what components it has, how you do build/estimate each component, how do you sample from it and, finally, how you evaluate the resulting synthetic dataset.

BIO: Bruno Dias is one of the data scientists within James R&D team. He has a background in computer science (intelligent systems & robotics) and has been working on machine learning explainability and missing data analysis/imputation. Also, he has experience with front-end development, audio signal processing,
and semantic web. Also, he has a big interest in healthcare informatics and computational creativity.

19.40 - 20.10: Break with drinks & snacks

20.10 - 20.20: Community pitches

20.20 - 21.00: “Graph neural networks for multirelational data” by Vadym Safronov

Running promotions in a retail chain, delivering organizational changes, or simulating clinical trials for a new drug - what do these activities have in common?

Network analysis, or reasoning with graphs (sets of pairwise relations) is a universal design language that can address all of the listed challenges. We will review data, processes, and interactions in complex networked systems, as well as some ways to model these in Python with PyTorch. Connections do matter, and we will go through evolution of methods addressing one specific problem - link prediction.

Would a combination of promotions kill sales, whether a team would perform or not, or would two drugs lead to a side effect - these problems can be solved by looking into structural roles entities play in their networks.

BIO: Vadym Safronov is a consultant in Customer Relations Management specializing in Machine Learning on graphs with Key Points.

21.00 - 21.30: Networking with drinks & snacks

Don’t hesitate to reach out if you have any questions or if you’d like to present at or host one of the future PyData’s.

See you soon!

Cheers,
The Jungle Crew