Master Assignment
Predicting young soccer players peak potential with optimal age
Type: Master CS
Location: Scisports
Period: Oct, 2018 - Mar, 2018
Student: Tahir, A.R. (Abdul Rehman, student M-CS)
Date final project: March 27, 2018
Supervisors:
Abstract:
Potential is the individual ability to perform. To measure the potential, experts systematically use many intelligence, ability, and competence based tests and makes a composite score from their results, which is considered to be an individual Potential. Most of the psychologists agree on the fact that potential increases and decreases with age. In this research, we developed an algorithm that can be used to predict the peak potential of young soccer players with optimal age. We used different machine learning techniques from traditional methods to deep learning methods to develop an algorithm that can predict the peak potential of young soccer players using their playing data between the age of 15 till 19. We used Lasso regression and FeedForward neural networks as our baseline models. We considered this problem as a time-series forecasting problem or sequence prediction problem. Our proposed model is a variant of recurrent neural networks– LSTMs. We have found that LSTMs outperformed baseline models and performed with zero prediction error on the test set when used with player-specific models.