Machine learning based Indian premier league (IPL) game predictions

• Main Authors: Mohmmad, Sallauddin, Raju, Oggula, Sridhar, Kankanala, Karivedula, Sheshipal, Laxmi Prasanna, Chindrala, Shabana
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Algorithms; Classifiers; Data analysis; Decision trees; Machine learning; Support vector machines
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0195894

Indian Premier League (IPL) is a famous Twenty-20 League conducted by the Board of Control for Cricket in India (BCCI). It was started in April 2008 and completed its fifteen seasons in 2022. The current, i.e., the fifteenth IPL season, was held in May 2022. IPL is a popular sport where it has a large set of the audience throughout the country. Therefore, every cricket fan would be eager to know and predict the IPL match results. This project is about a detailed exploratory data analysis of IPL matches conducted from the year 2008 till matches held in 2019. Here, we analyze the overall IPL match scores, best batting and bowling performances, the team with a more significant number of wins, the most successful IPL team, the most valuable players and their best performance range, and so on. The complete dataset is collected from officials of BCCI of IPL matches held from 2008-2019 through Kaggle. Here, the testing accuracy of SVM classifier is highest at 0.9159, and the next highest is the Decision Tree algorithm which gave 0.8225 accuracies. The second highest is Logistic Regression, which gave 0.8159 accuracies, and the Random Forest algorithm, with 0.7563 accuracies. As the SVM classifier has the highest accuracy of all the four models, we use that model to develop the analyzer model for the project.