Cluster-Centric Based Hybrid Approach for Cricket Sports Analytics Using Machine Learning

Sports Analytics is now playing a vital role in the sports industry playing a multifaceted role including decision making. Analytics in sports can be used by various stakeholders like the investors, sponsors, sports viewers and the team themselves. Analytics in cricket, being one of the most popular games of the world, pulls research interest in a greater way. Cricket related data includes statistics of players in the particular tournament or overall career statistics, team statistics, tournament statistics, etc. and the data may take a wide range related to the tournament or any particular team or any particular player of the team. Team’s performance is based on selection of players and hence the need for predicting the outcome of a match based on selected players becomes important. The proposed methodology uses a predictive machine learning model for predicting the outcome of the match based on the performance of a particular batsman or bowler. The increased impact of perfect selection adds a challenging aspect to the tournament which enhances the tournament’s popularity. This in turn will improve television contracts, fans store merchandise, sponsorship, ticket sales. The proposed methodology used various machine learning approaches such as Linear SVM, Logistic Regression, Decision Tree, KNN, and Gaussian Naive Bayes. A detailed comparison is made on the usage of these models. The proposed model is implemented using Python with Flask and is validated using the T20 International World Cup 2021 dataset. The results from all the machine learning models are compared and recorded. Logistic regression is proven to perform best out of all models to predict the winning probability of a cricket team.