Desarrollo de un sistema de predicción de jugadores titulares de un equipo de fútbol aplicando redes neuronales artificiales.

Abstract

Over time, data analytics of soccer players has increased to improve performance and decision making for championships, for this reason a prediction system was implemented by applying data science. Decision-making in soccer matches is complex when choosing players who play in the same positions, that is, the performance of each player must be detected in order to select the starting eleven and have a better chance of winning. This technological proposal is based on data science. The methodology used for the proposed DSS is known as KDD. Knowledge Discovery in Databases, which is a systematic, orderly and structured approach for the collection of large data sets, to then process them, obtain information and knowledge, to use them in various decisions and in any field, To achieve this, artificial neural networks were implemented to help with the automation of the predictions of the best starters. The results of this work are: The results of this work are: a prediction system of starting players of a soccer team by means of neural networks that helps in the automatic analysis of the performance of each match, the position that the player defends best, the pace at which he runs the field, the best throw. The tests showed acceptable and unacceptable results of the trained model, showing an accuracy of 42.86% in test A, with a smaller data, indicating that the Convolutional model (CNN) is regularly efficient, thus a second test B was performed, with a more extensive data showing an accuracy of 79.87%, indicating that the model is efficient. On the other hand, the results to sports and technology experts, offered us a positive critique accepting the hypothesis with an accuracy of 90%, in addition to obtaining the data from the use of the application.