PREDICTING BADMINTON MATCH RESULT USING LOGISTIC REGRESSION

• Main Author: Wellianto, Richard
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/51249
• Institution: Institut Teknologi Bandung

Indonesia participate in various types of international-level sport competitions. One of them is badminton, which is Indonesia’s best performing sport internationally. Maintaining that performance costs a lot, which can be relieved with financial support from sponsors. Sponsors prefer to support a capable player. The ability to predict player’s winning in future matches will convince sponsors of the player’s capabilities. Predicting a match is done by using variables affecting match result, e.g past matches’ score or head to head win percentage against that particular opponent. One of the method that can be used to predict the result of a badminton match is logistic regression. Logistic regression is a method used to predict binary variable, a variable which only take two values. Logistic regression used data from past matches to predict future matches’ result. Independent variable used in this project is percentage of points gained without the opponent’s fault, percentage of points given to the opponent through fault, and win percentage of the player against that particular opponent before the match. Training data used in this project is badminton match data from 2018 to 2019. Testing data used in this project is badminton match data at the end of 2019. A predictive software is built using logistic regression method to predict badminton match result. Evaluation of the method shows that the prediction accuracy rose when there are more past match data of the pairing used as test data. Evaluation shows that the prediction accuracy is 57.1%. This result is good enough considering there are no past experiment about predicting badminton match result.