A carregar...
Publicação
Spatiotemporal variability on triple jump : a case study of Patrícia Mamona with detrended fluctuation analysis and random forest analysis
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.37 MB | Adobe PDF |
Orientador(es)
Resumo(s)
Este estudo tem como objetivo compreender as adaptações da atleta Patricia Mamona, vice-campeã olímpica em Tóquio 2020, às diferentes fases do triplo salto e identificar os fatores determinantes para a sua performance, utilizando as técnicas avançadas de análise de dados. Com base em dados biomecânicos pré-existentes de saltos realizados entre 2015 e 2023, foram empregues as técnicas: (i) Detrended Fluctuation Analysis que permitiu estudar a variabilidade e persistências das velocidades e comprimentos de passos dos cinco últimos apoios. Esta análise longitudinal forneceu insights sobre as adaptações da atleta e sugeriu evoluções e adaptações significativas da técnica baseadas na velocidade e comprimento do Hop. (ii) Machine Learning, que através da interpretação do mean decrease in impurity do modelo Random Forest treinado, permitiu hierarquizar as variáveis que mais contribuem para explicar a distância efetiva de salto, tendo as variáveis relacionadas com o Hop emergido como as melhores preditoras da mesma. A interpretação conjunta das técnicas sugere que o Hop é uma fase critica do triplo salto e a capacidade de adaptação da atleta durante este momento deve ser trabalhada melhorando a adaptabilidade ao contexto. Diferentes velocidades de aproximação de tipos de superfície podem ser incluídos no treino com vista à otimização da performance.
This study aims to understand the adaptations of athlete Patricia Mamona, Olympic silver medallist in Tokyo 2020, to the different phases of the triple jump and identify the determining factors for her performance, using advanced data analysis techniques. Based on pre-existing biomechanical data from jumps performed between 2015 and 2023, the following techniques were employed: (i) Detrended Fluctuation Analysis, which allowed for the study of variability and persistence in velocities and step lengths of the last five supports. This longitudinal analysis provided insights into the athlete's adaptations and suggested significant evolutions and adaptations of technique based on the speed and length of the Hop. (ii) Machine Learning, which, through the interpretation of the mean decrease in impurity of the trained Random Forest model, allowed for the ranking of variables that most contribute to explaining the effective jump distance, with variables related to the Hop emerging as the best predictors. The joint interpretation of the techniques suggests that the Hop is a critical phase of the triple jump, and the athlete's ability to adapt during this moment should be worked on by improving adaptability to context. Different approach velocities and types of surfaces can be included in training to optimise performance .
This study aims to understand the adaptations of athlete Patricia Mamona, Olympic silver medallist in Tokyo 2020, to the different phases of the triple jump and identify the determining factors for her performance, using advanced data analysis techniques. Based on pre-existing biomechanical data from jumps performed between 2015 and 2023, the following techniques were employed: (i) Detrended Fluctuation Analysis, which allowed for the study of variability and persistence in velocities and step lengths of the last five supports. This longitudinal analysis provided insights into the athlete's adaptations and suggested significant evolutions and adaptations of technique based on the speed and length of the Hop. (ii) Machine Learning, which, through the interpretation of the mean decrease in impurity of the trained Random Forest model, allowed for the ranking of variables that most contribute to explaining the effective jump distance, with variables related to the Hop emerging as the best predictors. The joint interpretation of the techniques suggests that the Hop is a critical phase of the triple jump, and the athlete's ability to adapt during this moment should be worked on by improving adaptability to context. Different approach velocities and types of surfaces can be included in training to optimise performance .
Descrição
Palavras-chave
Análise da performance Atletas de alto-rendimento Atletismo Biomecânica Ciências do Desporto Cinemática Detrended fluctuation analysis Machine learning