Projecting Injury Risk Of Soccer Players In The Next Season By Leveraging Various Machine Learning Algorithms.

Football is such a popular sport on Earth that no other sport is capable of surpassing it with regard to audience, number of people playing it, and the enormous number of financial companies that revolve around this sport. However, players suffering from injuries are very frequent in soccer because of its high competitive nature and exhausting physical and mental needs. Soccer players suffer from various non-contract injuries (such as hamstring strain injuries), which impede their participation in the matches for a substantial amount of days, leading to poor performance of the team and financial losses to football clubs. These injuries are the consequence of the various internal agents (sleep hours, weight, physical fitness) and external components (training hours, number of matches per week, playing position), which make it difficult to recognize any factors as a sole contributor, making it difficult to identify a preventive technique to avoid injuries. To address this issue, machine learning is an appropriate tool to support the soccer team management in estimating the injury risk of players. This study intends to conduct a comparative analysis by applying different machine learning techniques (classification techniques: logistic algorithm, K-Nearest Neighbours, Support Vector Machine (SVM), regression techniques: gradient boosting and random forest, and naïve bayes algorithms) for predicting injury risk of university soccer players for the next season. This study will help the team management decide which technique is best for predicting injuries of players for the next season, and identify injury-prone players for taking preventive action for those players.