Outcome Prediction of ODI Cricket Matches using Decision Trees and MLP Networks

Abstract

Applications of machine learning supplemented with data mining techniques has become a hot topic for research worldwide, sports analytics is no exception though. Cricket is one of the most popular sports in Australia, Caribbean, UK and South Asian nations with a net fan base of around 2.5 billion. The game has tremendous spectator support in more than 100 nations and the masses show great interest in predicting the game outcomes. There are lots of pre-game and in-game attributes which decides the outcome of a cricket match. Pre-game attributes like the venue, past track-records, innings(first/second), team strength etc. and the various in-game attributes like toss, run rate, wickets remaining, strike rate etc. influence the result of a match in a predominant manner. In this study, 2 different ML approaches namely Decision Trees and Multilayer Perceptron Network have been used to analyse the effect produced on the outcome of a cricket match due to these varied factors. Based on these results CricAI: Cricket Match Outcome Prediction System has been developed. The designed tool takes into consideration the pregame attributes like the ground, venue (home, away, neutral) and innings (first/second) for predicting the final result of given match.