PyData London 2024

Generating embeddings for Yu-Gi-Oh Cards: A NumPy Approach to Represent Complex Data
06-16, 15:00–15:40 (Europe/London), Minories

Dive into the methodology of building a concurrence matrix to produce dense representations of cards, which pave the way for other, more interesting tasks such as recommendation systems. This session offers a blend of web scraping, clever coding practises and numpy's computational prowess, culminating in an illustrative card recommendation example.


In a world where data intersects with everything around us, trading card games like Yu-Gi-Oh! are not exempt from crossing paths with advanced data science techniques where the potential for innovation is immense. This talk at PyData focuses on the intersection of web-obtained data and NumPy to create an interesting approach for Yu-Gi-Oh! card recommendations.

The session begins by exploring the challenges of using complex data for data science purposes and how and why embeddings are a good solution for such a task. Then describes how is it possible to obtain a comprehensive dataset of Yu-Gi-Oh! cards.

The heart of the talk demonstrates the power of NumPy, a cornerstone library in the Python data science ecosystem, in generating dense representations of Yu-Gi-Oh! cards. By constructing a concurrence matrix that captures the intricate relationships between cards based on the expert knowledge of thousands of players, the talk showcases how NumPy can be harnessed to transform these relationships into meaningful embeddings. These embeddings serve as the foundation for a nuanced understanding of card dynamics and player preferences.

To top it off, there is a quick demonstration of an application of these embeddings in a real-world scenario—a card recommendation system. By leveraging the dense representations, the system can provide players with personalised card suggestions, potentially enhancing their gameplay and deck-building strategy.

Attendees will leave with a good understanding of what embeddings are, how is it possible to create embeddings with NumPy, and applying these techniques to recommendation systems. Whether you're a data scientist looking to apply your skills to new domains or a Yu-Gi-Oh! enthusiast interested in the analytical aspect of the game, this talk promises to offer fresh perspectives and actionable knowledge.
Join us me an engaging session that combines the thrill of one of the world's most popular trading card games with the cutting-edge capabilities of NumPy and machine learning.


Prior Knowledge Expected

No previous knowledge expected

I am an MLOps engineer at Simply Business; before that I worked as a data scientist for a market intelligence company and as a software developer creating web and mobile apps for insurance companies.

I hold a Masters Degree in Data Science and I also run a 17K subscriber YouTube channel about programming, machine learning and data science.