Big Data Fusion, Computer Analysis, and Benfica Dimension Optimization for Football Talent Selection

Aiming at the problem that multisource data are abundant but decision-making depends on experience and cost constraints are difficult to quantify in youth football training and selection, this paper proposes a data-driven decision support method for the selection process. Based on the six-dimensional ability framework, a multisource time series feature system is constructed by integrating competition events, physical fitness monitoring, and expert evaluation data and the proposed TabNet–long short-term memory model is used for potential prediction. The empirical results show that this method is superior to the traditional baseline model in terms of prediction performance and decision stability. Through threshold-cost joint analysis, the model output is embedded in selection decision-making and combined with the interpretation mechanism based on Shapley additive explanations, the intelligibility of the results is improved, and a universal decision support framework is provided for talent selection under resource constraints.