Data Mining in Sports: A Neural Network Approach

Abstract. Data Mining techniques have been applied successfully in many scientific, industrial and business domains. The area of professional sport is well known for the vast amounts of data collected for each player, training session, team, game and season, however the effective use of this data continues to be limited. Many sporting organisations have begun to realise that there is a wealth of untapped knowledge contained in their data and there is an increasing interest in techniques to utilize the data. The aim of this study is to investigate the potential of neural networks (NNs) to assist in the data mining process for the talent identification problem. Neural networks use a supervised learning approach, learning from training examples, adjusting weights to reduce the error between the correct result and the result produced by the network. They endeavour to determine a general relationship between the inputs and outputs provided. Once trained, neural networks can be used to predict outputs based on input data alone. The neural network approach will be applied to the selection of players in the annual Australian Football League (AFL) National Draft. Results from this study suggest that neural networks have the potential to assist recruiting managers in the talent identification process