Machine learning-based multi-model framework for football team optimization

Football is a sport that enjoys massive popularity and at the same time it is a huge financial market. The worth of football players in the market nowadays has a tendency of being on a constant rise. The use of computer science in sports analysis has totally transformed the process of assessing players and selecting teams by means of the sophistication in data analytics, machine learning (ML), and artificial intelligence (AI). The model that this study uncovers is a player recommendation model that about data analytics along with AI-based visualization techniques which help crisis and recruitment in football. This research is based on the FIFA-20 dataset, but it is not aimed at making predictions about video-game ratings. The dataset is instead considered to be a structured and standardized proxy of the actual football performance characteristics, with the rating being a summative assessment of technical, physical and tactical properties that are of interest when applied to football analytics. A model that utilizes a Multi-Layer Perceptron (MLP) architecture with two hidden layers is proposed and evaluated in comparison to the Rio de Janeiro and other models like Optimized Linear Regression, LightGBM, Random Forest Regression (RFR), and XGBoost. The suggested MLP architecture won in generalizability with an R² score of 99.13, RMSE of 0.6410, and MAE of 0.4485, which were the scores of competing models. The cross-validation findings, which provided an average R² and a standard deviation, further substantiated its stability. This research backs up the claim that AI-based tools, among other things, could be used in talent spotting with precious tips for player selection, efficiency evaluation, and team distribution in contemporary football.