CRICKET SCORE AND WINNER PREDICTOR
Cricket matches are known to be tremendously exciting but also, at times, extremely unpredictable. Players are in a constant state of training to emerge triumphant in their matches. To train their teams, coaches use previous performances of their respective teams to target areas where the team needs improvement. This would entail that coaches spend a lot of hours going through video footage trying to analyze what happened and what could have happened had their tactics been different. This wastes precious time and is a major cause of inefficiency in the work-flow. Resolving this would be of tremendous help to coaches as well as their teams and would give them an edge over other teams. This project aims to optimize this process of analyzing cricket matches to change tactics and encourage teams to perform better against certain rival teams through data mining algorithms. The goal is to create a model through the Linear Regression algorithm that predicts the score of an ongoing match by giving ball-to-ball data of previous similar matches (played on the same ground, played against the same team etc as the ongoing match) and determining the chances of positive outcomes for a particular team.
Team Members:
Keya Shukla (171210033) - Group Leader
Tanika Jindal (171210056)
Srijan Gupta (171210051)
Data Set Used:
https://cricsheet.org/
3. PROBLEM STATEMENT
Score prediction in cricket is of utmost importance.
It is used by coaches to train teams based on past performances and to formulate strategies for
future games.
Also instrumental to spectators in placing bets.
Prediction is generally done using a Regression algorithm.
Problem Statement : Create a model that predicts the number of runs, and consequently the
winner of a T20 match.
4. OVERVIEW
In this project, we have used Multiple Linear Regression to predict the number of runs made and
consequently the winner of a T20 match.
The data set used is : https://cricsheet.org/
Statistical technique that uses multiple features to predict the value of a response variable.
Models a linear relationship between independent and dependent variables.
The equation used is :
5. IMPLEMENTATION
The training and test data was split according to a 75-25 % ratio.
The model was applied to the data set to train it for accurate prediction.
Sklearn library was used and LinearRegression() function is used.
Predicted y values are found by predict() function, by giving X test values as input.
R squared value is found by using the score() function.
The accuracy is calculated by a custom accuracy function.
Accuracy is calculated by : ( predicted score – actual score ) , if this difference falls below a
particular threshold, it is considered as a correct prediction.
Accuracy obtained : 72.34 %
6. PREVIOUS WORKS
A research has been done in which classification algorithms such as Naive Bayes, Random Forest,
Multiclass SVM and Decision Tree classifiers are used.
Their approach was to classify number of runs and number of wickets in different ranges.
Another was done using K-Nearest Neighbours, by taking toss decision and venue of match
along with strength of team as features.
Lastly, a research was conducted to formulate CricAI by using AI techniques along with Bayesian
Classifiers.
7. CONCLUSION AND RESULTS
The prediction was carried out by using Linear Regression.
Such a prediction will help in efficient training and studying for future matches, as well as for
betting purposes.
Through comparison it was found that Random Forest Regression would give a higher accuracy
and thus gives scope for future research.
The Linear Regression accuracy could be improved by increasing the data set or by choosing
alternate features for prediction.
To conclude, with an accuracy of 72.34 %, the model is able to correctly predict the number of
runs. This gives an idea about the winner of the match.
8. REFERENCES
Passi, Kalpdrum & Pandey, Niravkumar. (2018). Increased Prediction Accuracy in the Game of
Cricket Using Machine Learning. International Journal of Data Mining & Knowledge Management
Process. 8. 19-36. 10.5121/ijdkp.2018.8203.
Predicting the Outcome of ODI Cricket Matches: A Team Composition Based Approach by
Madan Gopal Jhawar, Vikram Pudi in European Conference on Machine Learning and Principles
and Practice of Knowledge Discovery in Databases (ECML-PKDD 2016)
A. Kaluarachchi and S. V. Aparna, "CricAI: A classification based tool to predict the outcome in
ODI cricket," 2010 Fifth International Conference on Information and Automation for
Sustainability, Colombo, 2010, pp. 250-255.