Sports result prediction using data mining techniques in comparison with base line model

Sports prediction is one of the recent growing areas of interest entailing good prediction accuracy. Coaches require models in order to assess their players, analyse opponent teams and formulate winning strategies. Generation of comprehensive statistical data in sports has enabled data mining (DM) techniques to be applied to it in order to extract underlying predictive information. In this paper, an approach based on data mining is proposed for result prediction in sports. The work includes pre-processing of data, feature extraction, attribute selection and application of DM algorithms as a learning strategy. To validate our proposed model, a case study concerning prediction of the results of Indian Premier League (IPL) matches is illustrated. The constructed models are based on the performance of teams in past matches, player performance indices, opposition team information and external factors, and therefore, relevant features are engineered to indicate the same. The best prediction accuracy was found to be 70.58%.