Data-driven decisions in Sports

Abstract: In recent years, many sectors such as insurance, banking, retail, etc. have adopted Big Data architectures to boost their business activities. Such tools not only suppose a greater profit for thesecompanies but also allow them to gain a better understanding of their customers and their needs.These techniques are rapidly being adopted, this also being the case of sports and team sportsfor tasks such as injury prediction and prevention, performance improvement, or fan engagement.The aim of this project is to analyze the implications of data-driven decisions focusing on theiractual and future use in sports. Finally, a player scouting and team tailoring application would bedesigned and deployed to help the technical staff decision-making process which will also supposea budget optimization. For doing so, “Python” programming language and “Rapidminer” will beused, implementing “fuzzy logic” techniques for player scouting and “knapsack problem” algorithms for budget optimization plus an additional price prediction algorithm. The outcome wouldbe the application which given certain player needs (e.g., a midfielder with a high pass accuracyand a high ball recovery and a goalkeeper with a big number of saves and many minutes played)and the available budget will suggest the best possible combination of players given the availablebudget and the algorithm capable of predicting prices. This project also intends to study how thisapplication could be deployed in a real case situation by estimating the work team and budget todo so.