A professional rugby club.
In rugby, as in other sports, coaches prepare their teams for matches by analysing how the opposition plays and the tactics they use. Current analysis is carried out by watching television footage of each opposition’s past few games and noting the instances of game tactics and particular moves. Additional insight of varying usefulness can be gleaned from data provided by sports data companies. Faculty was asked to develop a data science approach to analysing team performance to help the club’s coaches prepare better for each game.
First, we built a database that included more than one million on-pitch actions. We augmented it with a dashboard with filter functionality that could plot where on the pitch certain moves and plays were made.
Second, we refined the model to help users to analyse different lineout routines (an important element in the game of rugby) and which of those were more successful than others. Such intelligence is very helpful for coaches when analysing opposition tactics and developing their own tactics. We used a gradient boosted model to predict the success of a particular routine based on its position on the touchline, the number of players involved in the lineout and the areas of the pitch the routine targeted.
The club now routinely uses the dashboard we developed as an integral part of its game preparation.
Screenshots of the dashboards, showing analysis of the data, created using Plotly Dash, supported by Faculty Platform.