Predictive performance analysis of players against training plan

Abstract

Performance of a player in competitive sports is collective result of skill, physical and mental fitness, diet, training etc. Every human being is different and therefore personalization is required in every aspect. The wide use of computer system in different aspects of training and coaching makes it easy to generate and gather data in digital form. A powerful tool is required for analysis and interpretation of this data. The knowledge extracted from such data could be helpful in decision-making, system learning and automation. The training is requires to be individualized. This individualization helps to achieve maximum performance from each individual player. To ensure personalization it requires individual monitoring and evaluation, which is quite impossible without use of any tool. A system ‘Swimming Coach Assistant’ developed for predictive performance analysis of a player and assist coaches to extend personalized coaching to the player. The anthropometric measurements as suggested by the Heath-Carter method of Somatotyping of the player are used to describe the present morphological conformation of the player the nearest somatotypes of the players readily available in the system database are identify using distance formula (Somatotype Dispersion Distance). System suggests the best training plan previously identified and recorded by using data clustering approach. Otherwise, coach assign initial training plan based on his knowledge and expertise. The system provides a plot of performance of players in the practice session for the assigned training plan. A time series approach is use for fitting a straight line for the gathered performance data. This result provides a valuable feedback to the coaches to individualize the training activity and can predict the future performance if same training plan continues.