An Artificial Neural Network Approach for Classifying Cricket Batsman's Performance by Adam Optimizer and Prediction by Derived Attributes

In a cricket match, for one to win it is necessary that the team management selects the best starting 11. To do this a lot of attributes need to be considered like the averages, strike rate and other statistical data. But it is also necessary to take into consideration the non-statistical data like the form of a player, whether or not a team is under pressure and other such factors. In this paper we have considered all of these factors to predict a batsman's performance. We have proposed an Artificial Neural Network to predict the players performance using t20 dataset. Mathematical models for predicting the batting form and bowling form of the players are formulated and they are used as attributes in the proposed algorithm. The prediction accuracy of the proposed artificial neural network has been compared with that of machine learning models. Data is provided containing all the aggregate values of that attribute till that match to predict the output. The output is classified into five categories where 1 is the lowest and 5 being the best performance. This model has gained better accuracy compared to the accuracy obtained by Random Forest, KNN and SVM classifiers. It would help the team management select a better team to accomplish higher probability of winning the matches and cash prize awards.