Desarrollo de un modelo predictivo para el análisis de anomalías de desempeño en futbolistas profesionales utilizando minería de opiniones.

Abstract

Nowadays, soccer requires tools to confirm through statistics when a player’s performance is dropping, and consequently, propose corrective. The present project aims to accomplish two goals. The first one focuses on establishing a mechanism to identify anomalies in the performance of professional soccer players. We understand as an anomaly, the development (in terms of performance and financial valuation) of an athlete who has not fulfilled his projection as a future star. Secondly, based on the identification of these anomalous performances, we intend to develop a model that allows explaining these behaviours through opinion mining from what has been said about the athlete in Twitter. The relevant information will be extracted from people’s tweets and analysed through Sentiment Analysis, in addition to word cleaning and vectorization processes. To do so, the CRISP-DM methodology will be followed, in addition to information extraction processes, execution of Python libraries for text processing, among other tools; all for implementing methods of Artificial Intelligence and Machine Learning. At the end, it was possible to relate people’s opinions with players’ performance to predict if a player is risky and to support the decision-making process about the player’s future within the soccer team.