Applying Data Science for Cricket Predictions

Abstract

In any sport selecting the best players is an important and crucial task. In the game of Cricket, the performance of the players depends on various factors such as the team they are playing against, venue of the match etc. From 15 to 20 players, cricket management team, coach and captain select 11 players for the match. They select them by analyzing and using various statistics and strategies. Every batsman contributes to the game by scoring maximum runs possible and each bowler contributes by taking maximum wickets and avoid the opponent team from scoring runs. This paper predicts the performance of players such as strike-rate of the batsman, number of wickets a bowler can take and his economy. These predictions are regression problems and various machine learning regression algorithms have been used to predict the statistics of each player. We used random forest, decision tree, linear regression, KNN, Gradient boosting regressors to generate the prediction models for the problems. For each statistic prediction certain regression algorithms are found to be accurate and effective and are discussed further in this paper.