Predicting a T20 cricket match result while the match is in progress

Abstract

There has been research done on ODI and Test match cricket but very few on T20 cricket, which is currently more favorite than its older brothers. And that’s why we decided to do research on this format of the game. The result of a T20 cricket match depends on lots of in game and pre-game attributes. Pre-game attributes like condition, venue, pitch, team strength etc. and in game attributes like wickets in hand, run rate, total run, strike rate etc. influence a match result predominantly. We gave more emphasis on in game attributes as our prediction will be when match is in progress. Our intentions would be to finding out the attributes which is most affecting the result in different phases of the game. We broke an innings into three phases: Power-play (1-6 overs), Mid-overs (7- 16) and final overs (17-20). Prediction will be active till the last over of mid overs phase. We consider an entire cycle of process of data mining, decision making and preparing a model to predict. Mining the data according to the attributes and different phases we have divided important to construct meaningful statistics. Modeling a problem for prediction requires several intelligent assumptions and molding the problem with collected data-sets. As we already mentioned cricket is a game of uncertainty and T20 format is the most unpredictable format rather than the other two format because it is the shortest format of the game and one over can change the result of a game. In this project we tried to design a prediction model which can go with this unpredictability and try to make a result prediction.