In-game behaviour analysis of football players using machine learning techniques based on player statistics

Abstract

The purpose of this research was to determine the on-field playing positions of a group of football players based on their technical-tactical behaviour using machine learning algorithms. Each player was characterized according to a set of 52 non-spatiotemporal descriptors including offensive, defensive and build-up variables that were computed from OPTA’s on-ball event records of the matches for 18 national leagues between the 2012 and 2019 seasons. To test whether positions could be identified from the statistical performance of the players, the dimensionality reduction techniques were used. To better understand the differences between the player positions, the most discriminatory variables for each group were obtained as a set of rules discovered by RIPPER, a machine learning algorithm. From the combination of both techniques, we obtained useful conclusions to enhance the performance of players and to identify positions on the field. The study demonstrates the suitability and potential of artificial intelligence to characterize players' positions according to their technical-tactical behaviour, providing valuable information to the professionals of this sport.

Machine learning application in soccer: a systematic review

“…The current trend focusing on talent has defined which term as a dynamically varying relationship molded by the constraints imposed by the physical and social environments, the task experienced, and the personal resources of a player [ 50 ], highlighting talent identification in one of the more important challenges in soccer, not only during early ages [ 8 , 10 ], but also for player transferal between soccer clubs [ 9 ]. In addition, ML can be used in this way to predict the most suitable players for decision-making for competition starter players (choosing players and playing positions) [ 10 , 33 ].…”

Application of Distributed Probability Model in Sports Based on Deep Learning: Deep Belief Network (DL-DBN) Algorithm for Human Behaviour Analysis

“…Step 3. 1/ 􏽐 n j�1 e θ L j s (i) normalizes the probabilistic model so that the θ L 1 total probability is the one and the decision variables for the system is shown in equation (3).…”

“…Furthermore, an individual's sports behaviour complete representation is designed to support specific spatiotemporal options. Crowd intelligence mechanism aids in reviewing and players' future results are predicted and they are fingered in a National Basketball Association league All-Star game [3]. e goal of the study is to analyse the soccer player's behaviour using machine learning algorithms with various classifying parameters that are derived from OPTA's (Optimum Performance eoretically Attainable) on-ball event [4].…”

Football Match Line-Up Prediction Based on Physiological Variables: A Machine Learning Approach

“…Artificial Intelligence (AI) presence in sports is gradually increasing, in particular in football, where ML algorithms are used to detect meaningful patterns based on positional data [6]. ML is already used in football to predict and prevent injuries in players [5,9,10], as well as in the categorization of football players and football training sessions [1,11], in the evaluation of football players regarding their market value [12,13], and in predicting the results of football matches [4,14], among others.…”

“…In terms of categorization, there are also some important studies. In [11] authors aimed at identifying technical-tactical behaviours of players using their statistics, without including spatial-temporal descriptors by using ML methods. In this work, authors assessed the capability of ML to identify the most influential variables for each of the positions on the field and to find groups of outlier players.…”