Prediction Of Outcomes Of Extra Deliveries In T-20I Cricket By Using Regression And Various Machine Learning Models

Abstract

T20 cricket is a modern-day popular, fast-paced, and highly competitive format where each team has 20 overs to score as many runs as possible. Extra runs from no-balls, wides, byes, and leg byes can significantly influence the outcome of a match. Reducing these extras is vital for teams aiming to maintain control and increase their chances of winning. This study intends to investigate whether extra deliveries in an over have a significant effect on the result of a T-20 cricket match. To determine which factors, have the greatest impact and how these extra deliveries affect the outcome of the match, the study will look at a number of different variables. The aim of this study is to verify the commonly belief that more extra deliveries significantly reduce a team's chances of winnings. The study uses a classical logistic regression model and machine learning models including neural networks, XGBoost, decision trees, and k-nearest neighbors to evaluate the data from the last three Twenty20 International Cricket World Cup matches in order to finding the key factors that affect the outcome of the match. To improve the predicted accuracy of the model, the study takes into account variables, such the number of extras bowled (NEB), runs scored from extras (RSE), total score in first innings (TS1), total score in second innings (TS2) overs bowled (EB), number of wickets taken (NOW), and game winner (GW).