Cricket Team Prediction with Hadoop: Statistical Modeling Approach

Cricket is one of the most popular team games in the world. In this paper, we embark on predicting the best suitable Team to be lined for a particular match. We propose statistical modeling approach to predict the perfect players for the match to be played. As cricket is not a very simple sport, there are many factors affecting the line-up and selection of players for a particular match such as Player's Overall Stats, Player Performances with different Teams and the most important Last 5 Performances. All these factors have been considered for selection of players in playing 11 from the Team of 16. This work suggests that the relative team strength between the competing teams forms a distinctive feature for predicting the winner. Modeling the team strength boils down to modeling individual player batting and bowling performances, forming the basis of approach used. Career statistics as well as the recent performances of a player have been used to model. Player independent factors have also been considered in order to predict the outcome of a match. Experimental analysis was performed using Hadoop and Hive for Indian players. Results establish that proposed approach is able to obtain up to 91% accuracy as compared to the real results available over WWW.