Methods and challenges of using predictive analytics in sport: Winning one-dayinternational cricket matches
Loading...
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Sage Publications Ltd.
Abstract
The case study predicts match results for one-day international cricket played for fifty overs a side. The case is based on empirical research carried out on the actual data set of four major national sides (i.e., Australia, England, India, and Pakistan) with a total of 3424 official matches of a total of 8384 played between January 5th, 1971 and July 11th, 2019. We discuss the binary logistic regression method to predict the odds of a win (as the dependent variable) modeled on the independent variables classified across batting, bowling, fielding, and others. Overall, nine independent variables were considered – fours and sixes scored under batting, bowling economy, and extras conceded under bowling, fielding dismissals under fielding, and four generic variables (i.e., the season of play, pitch condition, umpires nationality, and debutants on each side) under others category. It may be noted that data was collected for the playing teams on all the batting, bowling, and fielding variables, as well as the number of debutants. Operationalization of the variables, assumptions considered for operationalization given the longitudinal spread of the data for forty-eight years, spatial spread with the involvement of multiple continents, and playing conditions were some challenges faced in the research design and data collection. The case will allow the readers to develop predictive models with a binary decision variable and large data sets. The readers will get an experience of not just developing complex predictive models but also their validation through in-sample and out-of-sample tests.
Description
Keywords
Citation
Sarangi, S. (2023).Methods and challenges of using predictive analytics in sport: Winning one-dayinternational cricket matches. Sage Publications Ltd. https://doi.org/10.4135/9781529628883