Football Analytics for Goal Prediction to Assess Player Performance

Abstract

Machine learning techniques are often used for sports analytics, such as player health prediction and avoidance, appraisal of prospective skill or market worth, and predicting team or player performance. This reshapes the sports performance and helps in coaching the teams and individuals. This research focuses on football analytics, which can help football managers and coaches for reshaping the performance of players to target the goal with higher accuracy and precision. The match results depend on the successful number of goals; any minor mistake may lead to failure. Other statistics, like shots on target and game possessions, have been gaining popularity in recent years. Several attributes are utilized to train an anticipated goal model formed by monitoring football data to evaluate the chance of a shot being a goal. Using historical data and advanced analytics, a credible prediction of a goal, as well as player and team performance, can be deduced. Furthermore, we address the identification and recording of personal talents and statistical categories that distinguish an exceptional goal scorer from the worst goal scorer through football analytics. Feature selection, data size, and parameters used may impact the results of the model. Our research proposes a Goal Prediction Model (GPM) with player analysis trained on data from 9,074 games, including 941,009 events from Europe's top 5 leagues containing the information of five seasons. Our model will explain the observations on expected goals through football analytics and monitor the performance of the players with respect to anticipated goals. This research could benefit football team managers and coaches by reshaping the performance of players.