Prediction of badminton match result using machine learning

Badminton is a sport that includes singles and doubles categories that play until 21 points. The Badminton World Federation (BWF) organises 26 tournaments a year, with competition divided into five levels. Due to that, a huge amount of data is collected every year and is available from various sources. Thus, this paper proposes to make a prediction of badminton match outcomes using a machine learning algorithm. A data set was extracted from the Kaggle website in a comma-separated values (CSV) file, and an ETL (Extract, Transform, Load) process was implemented before the data was loaded into the Apache Hive data warehouse. A new method of supervised machine learning is suggested to forecast the results of badminton matches by utilizing in-game statistics. In the past, badminton has not received as much attention in terms of outcome prediction compared to other sports. The study outlines techniques to gather eight specific features from publicly accessible match results presented in a standardized format. By applying logistic regression and K-nearest neighbor algorithms to these features, analysis is conducted on 14,722 professional-level matches that took place from 2018 to 2021. The accuracy of match outcome prediction was achieved between 76.91 % and 85.39%.