Titulo Estágio
Predicting Sports Players Performance
Áreas de especialidade
Sistemas Inteligentes
Local do Estágio
CISUC/TulaLabs
Enquadramento
Basketball performance depends on a combination of physical, functional and behavioral characteristics and sport-specific skills. This sport is highly exclusive, focused on an exceptional minority of individuals. The overwhelming majority of youth who participate in sport will not achieve an expert level and the paths for the development of young players towards adult basketball expert level are unclear. In general, development and promotion of young athletes is based on talent identification and development programmes. Substantial investments by Federations and professional clubs are made for early identification and promotion of “young gifted athletes”, and international competition is used as a benchmark of young elite performance. In basketball, game-related statistics in competition are being used to identify variables that can distinguish between successful teams, in both adults; and youth teams. However, the issue of individual performance of basketball players in game context has not been considered. Playing efficiency (game-related) of individual basketball players can be measured objectively with statistical records of the game, and may be used by coaches and sport scientists to distinguish player's sport-specific skills capabilities.
Objetivo
- Create a dataset based on the online available data
- Deal with a large quantity of data and multiple independent variables and interaction with different levels of variation within and between yearly-observations.
- Track super-elite performers since the first international youth competition and try to establish a relation between how much elite youth performance predicts adult super-elite performance.
- Investigate whether the findings of this work are relevant and can be applied to other (team) sports, such as football or volleyball
Plano de Trabalhos - Semestre 1
T1 - Analyse the State of the Art and increase familiarisation with the problem.
T2 - Construction of the dataset based on the online data.
T3 - Develop and proposal of the solution and of the validation strategies.
Write the Dissertation Plan.
Plano de Trabalhos - Semestre 2
T1 - Data Preprocessing, which includes data preparation, descriptive summarisation of the data, data cleaning (missing values, noise detection, outlier detection), data transformation (scaling, discretisation), data reduction (feature selection, feature extraction)
T2 - Find hidden patterns in the data, including visualisation of feature correlations, and low dimensional analysis of the dataset (PCA analysis)
T3 - Application of unsupervised learning methods (partition methods for clustering, hierarchical clustering methods
T4 - Validation of the results according to the previously defined strategies.
T5 - Write a scientific publication.
T6 - Write the Dissertation.
Condições
The work is to be conducted at the CMS and ECOS laboratories of CISUC, where it will be a regularly monitored by the supervisors. A workplace will be provided as well as the required computational resources.
Observações
This work will be developed in a partnership between CISUC and the Faculty of Sport Sciences and Physical Education.
Orientador
Nuno Lourenço/Hugo Oliveira
naml@dei.uc.pt 📩