Utilizing Machine Learning for Comprehensive Analysis and Predictive Modelling of IPL-T20 Cricket Matches

Abstract

Objective : Current study intends to develop a predictive model for Indian Premier League (IPL) cricket match results using machine learning techniques. In order to provide a precise framework that allows for the prediction of IPL match outcomes, it aims to examine player statistics, match dynamics, and historical data. Method : SVM, Random Forest, Logistic Regression, Decision Tree, and KNN models were used in this study to predict player performance on any given day. Form, fitness, and previous results were among the historical player data that were used as characteristics. Each model preceded through training and testing phases, with accuracy, precision, and recall metrics evaluated to determine the most effective algorithm for forecasting player performance. Findings : Final studies indicated that relative team strength of competitor teams, recent form of players, and opponent pairings are distinguishing features for predicting the performance of both players and teams on any given day. The multi-machine learning approach-based model that was constructed demonstrated an accuracy of 0.71, further indicating improved performance for the given challenge. Modelling team strength is similar to modelling individual player batting and bowling performances, which is the cornerstone of our approach. Novelty : This paper was designed based on a novel approach leveraging combinatorial machine learning methods. This has been found to demonstrate unprecedented performance improvement in predicting a player’s performance on a given day. Additionally, the presented approach may prove valuable in opening new avenues to advance machine learning applications in sports analytics by addressing the limitations of existing methods. Keywords: Machine Learning, Sports analytics, SVM, Random Forest, KNN