Smart Cricket team selection for Pakistan cricket team using Machine Learning & IoT

Abstract

Cricket is a renowned and most popular game not only in Pakistan but in the world as well. The passion of cricket in Pakistan is to the extent of obsession. The success of the team depends on the best selection of the players and their performances. It is a difficult challenge to find the best players, but it’s worth at the end if there is a correct selection of players. Many players complain that the team selection process is not on merit and it is a biased process. The former national selection committee made several changes to the T20 team during the last year or so, saying it was testing different combinations to pick the final line-up for the World T20. However, it had a major impact on Pakistan’s economy. A novel Machine learning based Cricket team selection technique is proposed and the complete processing of previous performance results would be done by using machine learning techniques to get top XI players to guarantee the success of that selected team. To implement the proposed technique, we manually collected the dataset from the PCB official website, and this PSL Dataset is then used in this research work. We used feature Extraction to extract the most useful features. Finally, we used different prediction/ regression models, i.e., Linear Regression, Random forest, and Decision tree, to predict the players’ performance in the next match. For batsmen performance prediction, Linear Regression gives 97.33% accuracy, Random Forest gives 95.80%, and Decision Tree gives 97.33% on average. For bowler’s performance prediction Linear Regression gives 98.38% accuracy, Random Forest gives 93.40%, and Decision Tree gives 98.64% on average. Our proposed model could easily be used for the team selection of the other cricket formats as well. Moreover, with little modification, the Team selection method could be easily used for the selection procedure of any sport.