An efficient team prediction for one day international matches using a hybrid approach of CS-PSO and machine learning algorithms

Abstract

Player classification is vital in cricket since it assists the coach and skipper in determining individual players' roles in the squad and allocating tasks appropriately. The performance statistics help to classify players as batsmen, bowlers, batting all-rounder, bowling all-rounder, and wicketkeeper. This research aims to correctly identify cricket teams in the one-day international format by categorizing players into five groups. Based on their previous and current performance, the players are rated as excellent, very good, good, satisfactory, or poor. An enhanced model for the game of cricket is presented in this study, in which an eleven-member team picked using an unbiased technique. Players should be selected based on their performance, batting average, bowling average, opposing team strength and weakness, etc. Nature-inspired algorithms are used for feature optimization to improve the accuracy of machine learning prediction models. The blending of Cuckoo Search and Particle Swarm Optimization is performed called CS-PSO, which successfully integrates the capabilities from both approaches to create reliable and suitable solutions in accomplishing global optimization efficiently. Using a hybrid of CS-PSO feature optimization and Support Vector Machine, batters, bowlers, batting all-rounders, bowling all-rounders, and wicketkeepers were picked with an accuracy of 97.14%, 97.04%, 97.28%, 97.29%, and 92.63%, respectively.