Dimensionality Reduction and Latent Space Modeling of Soccer Players’ Tactical Profiles

This thesis aims to develop an innovative methodology for analyzing and representing the tactical profiles of Serie A football players using detailed match-by-match statistical data. The primary objective is to delineate player behavior through the generation of various types of heatmaps that capture different types of actions. These heatmaps will be reduced in dimensionality, projecting playing styles into vector form, using two different approaches: a Variational Autoencoder and a Non-Negative Matrix Factorization. Once the characterizing vectors are extracted, they will be evaluated by a similarity analysis and tested with a proposal clustering pipeline, in order to detect patterns between players tactical behaviours.