Last Thursday, adsquare hosted the PyData meetup “PyData goes graphs”. Our data scientist Matthias Kirmse took this opportunity to give some insights into the data science projects we’re working on here. Matthias talked about the big data architecture for storing and processing big amounts of RTB bid requests and python tools like Pandas used for structuring and analyzing the data. He also talked about interesting problems solved by the team, for example the analyzing and filtering of irrelevant geo location data and coverage optimization for audience live product.

Romain Guigourès presented theoretical framework for analyzing graph structure. Based on basic concepts of information theory, he introduced an exploratory data analysis framework aiming at tracking structures in large graphs (graph clustering), visualizing the interactions within the graph, focusing on the most interesting patterns and finding the most representative items of a group. To illustrate the efficiency of the framework, he presented a detailed analysis on mobile phone data recorded in Ivory Coast over 5 months.

Finally Sirin Odrowski presented python-igraph, an interface to the igraph library, which provides functionality to analyze and visualize complex networks. She made a tutorial-style introduction to making graph plots with python-igraph.

As a big data company we love hosting these events, thanks to everyone who was there! Follow @adsquareTech to get updates on the upcoming events at adsquare!

Back to the blog