Machine Learning-Based Performance Prediction Models for Test Cricketers: A Data-Driven Approach

A potential strategy in the sports sector is to forecast of players' test cricket efficiency using ML methods. The opponent team, the field, the player's current form, etc. All have an impact on how well the players perform. Each batsman assists by scoring as many runs as they can, every bowler serves by taking the most wickets and allowing the fewest runs to be scored, and each all-rounder serves by scoring the most runs and taking the most wickets. The topic of this paper aims to anticipate player performance, especially the number of runs each batter will score, the number of wickets each bowler will take, and the usefulness of an all-rounder. The categorization concerns in each of the three departments, where variations for the number of runs and wickets vary, are meant to be resolved. This abstract gives a summary of the techniques of ML in predicting player performance, highlighting the methodologies, datasets, and benefits associated with this data-driven approach. By leveraging historical performance data of players and their attributes, machine learning models learn patterns and relationships to generate accurate predictions. We have used “orange” platform to create our prediction model in which we have used three different datasets of batsmen, bowlers and allrounders respectively. To create a prediction model for each of the three departments, we employed naive bayes, random forest, KNN, and tree models. For all three departments, the approach with the greatest accuracy was found to be the random forest algorithm with more than 96 percent of classification accuracy.