Intelligent decision analytics with recursive learning for competitive sports management

Abstract

The increasing availability of sports performance data has created new opportunities for data-supported decision-making in competitive sports management. However, redundant and strongly correlated performance indicators can increase analytical complexity and distort the influence of criteria in subsequent weighting and ranking procedures. This study proposes an intelligent decision-analytics framework that integrates recursive unsupervised feature selection, logarithmic percentage change-driven objective weighting (LOPCOW), and the preferences relative ordering by frequency inclusion technique (PROFIT). Descriptive analysis, correlation modelling, and hierarchical clustering are first applied to examine the distributional and structural relationships among the performance indicators. The recursive feature-selection procedure then evaluates each indicator according to its entropy-derived information content and correlation-based redundancy. Indicators providing limited additional information or substantial overlap are progressively removed while preserving at least 95% of the measured information and maintaining representation from each identified feature cluster. The retained indicators are objectively weighted using LOPCOW, and the simulated sports management alternatives are ranked using PROFIT. The framework is examined through a MATLAB-based case study constructed from simulated evaluation data rather than observations collected from real athletes, teams, or sports organizations. Under the specified Monte Carlo perturbation design, the same highest-ranked alternative was obtained in 9823 of 10,000 simulation runs, corresponding to a selection frequency of 98.23%. This result indicates computational ranking stability within the simulated experimental setting and should not be interpreted as evidence of practical reliability or improved real-world decision quality. The proposed framework provides a structured procedure for reducing redundant indicators, determining objective criterion weights, and ranking alternatives under controlled decision conditions. Further validation using empirical sports performance data, domain-expert assessments, and operational sports management settings is required to establish its practical effectiveness and generalizability.