Pattern matching in football event data

Video by Koen Vossen, founder of http://PySport.org - @PySportOrg
Part 1 of 3.

Video analysts spend a lot of time trawling through footage to find examples of situations they or the coach want to show the players. It’s tedious work and can take a long time. Time they could spend on other tasks such as the optimisation of communication to players and coaches.

In this series of tutorials we will look into possibilities to automated (parts of) the process using pattern matching.

Part 1: Are there already frameworks that can help us search for patterns, and can we apply the framework to event data? The answer is yes! One of the ways to search for patterns is using regular expressions. Whats a regular expression? Well we have all used them! The ‘Find and replace’ tool searches for a precise sequence of characters we use them intuitively without thought. Well imagine if instead of searching for a sequence of characters we search for a precise sequence of football actions. In this tutorial we’ll look into a way to encode event data into such a format that we can use regular expressions to it. Next step is to show the results in a pandas dataframe, or event plot it using a library like mplsoccer.

Part 2 + 3: Coming soon!

A full Python notebook for this tutorial can be found here: https://github.com/koenvo/kloppy/blob...