K Means clustering and descriptive analytics based performance recommending system for Kabaddi team and player

Abstract

In the contemporary era, Kabaddi is considered as a commercial game. Manual analytics were used in old times in which the best team or player was selected based on the background. However, in the present era, statistical analysis and machine learning are used in place of conventional methods of sports analytics in commercial sports such as Cricket, Football, etc. This study intends to analyze correlations for features of accumulated online datasets and to make predictions for the team and player performance using correlated features by using statistical machine learning approaches. The suggested methodology includes feature extraction, correlation identification, and implementing appropriate machine learning approaches. Initially, the parameters of the Kabaddi game concerns such as the impact of tosses, cards, and home ground on results were analysed. Subsequently, based on team and player data correlation and cluster analysis, important characteristics were identified and appropriate rank-based scoring was established. Finally, the regression-based prediction was suggested with an r2 score and a cross-validation score greater than 0.91 with the least errors. Finally, a trained machine-learning model with greater outcomes was suggested by verifying the parameters that were analysed. After the completion of analysis, the proposed techniques would be utilized in real-time scenarios using visual dashboards, deep learning, the Internet of Things, etc.