Quais ações são significativas para uma vitória no futebol? Uma análise por regressão logística
Carregando...
Data
2021-06-10
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O futebol é um esporte que envolve uma gama de variáveis físicas, técnicas e táticas. Apesar do aumento de estudos sobre a importância das variáveis técnicas no desempenho do jogo nos últimos anos, ainda o estudo das variáveis físicas tem sobressaído. Além disso, a maioria dos estudos tem utilizado análises dos dados mais simples, que muitas vezes não conseguem responder precisamente algumas questões relacionadas ao desempenho esportivo. Análises estatísticas mais
aprofundadas são importantes para melhor explorar as variáveis de desempenho e apresentar uma análise mais objetiva do jogo. Desta forma, o presente estudo teve a seguinte questão a ser respondida: é possível encontrar um modelo de
características técnicas individuais de acordo com as posições de jogo que são determinantes para vencer no futebol? Assim, o objetivo deste trabalho foi analisar a influência das ações técnicas individuais de cada posição de jogo sobre o resultado da partida através de um modelo de regressão logística. Foram analisados 5396 jogadores das 10 principais ligas nacionais (Alemanha, Argentina, Brasil, Colômbia,
Espanha, França, Holanda, Inglaterra, Itália e Portugal) nas partidas em que atuaram entre os anos de 2015 até 2019. Os indicadores foram obtidos através do site da Wyscout®, e então tratados em R® com o objetivo de compilar os dados em uma
única planilha de Excel. Esses dados foram então importados para o Phyton®, onde foram manipulados para classificá-los de acordo com a posição de jogo. Finalmente, por meio de técnicas de scikit-learning, foi realizado um modelo estatístico de regressão logística para responder à pergunta do estudo. Os resultados revelaram individualidades entre cada posição, dentro de cada liga, sobre as ações que foram determinantes para obter vitórias. Os mais determinantes, e com grande recorrência foram: número de gols, assistências e passes certos. Outro fator comum na influência positiva para o resultado foi o local da partida, de modo que ser mandante
demonstrou uma substancial vantagem. A partir dos resultados aqui apresentados treinadores poderiam utilizar as informações para ajustar treinamentos ou mesmo na busca por reforços específicos, de acordo com as ações significativas para a posição desejada.
Football is a sport that involves a range of physical, technical and tactical variables. Despite the increase in studies on the importance of technical variables in game performance in recent years, the study of physical variables still stands out. In addition, most studies have used simpler data analysis, which often fail to precisely answer some questions related to sports performance. Deeper statistical analysis is important to better explore performance variables and present a more objective analysis of the game. Thus, the present study had the following question to be answered: is it possible to find a model of individual technical characteristics according to the game positions that are decisive for winning in soccer? Thus, the objective of this work was to analyze the influence of the individual technical actions of each game position on the match result through a logistic regression model. A total of 5396 players from the top 10 national leagues (Germany, Argentina, Brazil, Colombia, Spain, France, Netherlands, England, Italy and Portugal) were analyzed in the matches they played from 2015 to 2019. The indicators were obtained from the website from Wyscout®, and then processed in R® with the aim of compiling the data in a single Excel spreadsheet. This data was then imported into Phyton®, where it was manipulated to classify it according to game position. Finally, using scikit learning techniques, a statistical logistic regression model was performed to answer the study question. The results revealed individualities between each position, within each league, on the actions that were decisive for obtaining victories. The most decisive, and with great recurrence were: number of goals, assists and successful passes. Another common factor in positively influencing the result was the match location, so, playing as the home team demonstrated a substantial advantage. Based on the results presented here, coaches could use the information to adjust training or even in the search for specific reinforcements, according to the significant actions for the desired position.
Football is a sport that involves a range of physical, technical and tactical variables. Despite the increase in studies on the importance of technical variables in game performance in recent years, the study of physical variables still stands out. In addition, most studies have used simpler data analysis, which often fail to precisely answer some questions related to sports performance. Deeper statistical analysis is important to better explore performance variables and present a more objective analysis of the game. Thus, the present study had the following question to be answered: is it possible to find a model of individual technical characteristics according to the game positions that are decisive for winning in soccer? Thus, the objective of this work was to analyze the influence of the individual technical actions of each game position on the match result through a logistic regression model. A total of 5396 players from the top 10 national leagues (Germany, Argentina, Brazil, Colombia, Spain, France, Netherlands, England, Italy and Portugal) were analyzed in the matches they played from 2015 to 2019. The indicators were obtained from the website from Wyscout®, and then processed in R® with the aim of compiling the data in a single Excel spreadsheet. This data was then imported into Phyton®, where it was manipulated to classify it according to game position. Finally, using scikit learning techniques, a statistical logistic regression model was performed to answer the study question. The results revealed individualities between each position, within each league, on the actions that were decisive for obtaining victories. The most decisive, and with great recurrence were: number of goals, assists and successful passes. Another common factor in positively influencing the result was the match location, so, playing as the home team demonstrated a substantial advantage. Based on the results presented here, coaches could use the information to adjust training or even in the search for specific reinforcements, according to the significant actions for the desired position.
Descrição
Palavras-chave
Football, Technical skill, Individual actions, Game position, Logistic regression, Futebol, Habilidade técnica, Ações individuais, Posição de jogo, Regressão logística, Análise de regressão