Dynamic Bradley-Terry 모델을 적용한 한국 프로 남자 배구팀 경기력 분석 (Performance Analysis of Korean Professional Man’s Volleyball Team applying Dynamic Bradley-Terry Model)

서지정보

· 발행기관 : 한국체육과학회
· 수록지 정보 : 한국체육과학회지 / 33권 / 1호 / 619 ~ 629페이지
· 저자명 : 허종관, 이봉근

초록

The purpose of this study is to estimate changes in the performance of the Korean professional volleyball team by applying the Dynamic Bradley-Terry model to Korea's volleyball game result records and compare the estimated results with the Static Bradley-Terry model. For the purpose of this study, the match result records of 252 games of 7 professional volleyball teams participating in the 2022-2023 V-League regular season were collected from the website of the Korea Volleyball Federation (KOVO) (https://www.kovo.co.kr/). The data was collected using a web crawling program created by the researcher using Python (python 3.11.2) program. For the collected data, the static Bradley-Terry model, which simply considered the opponent team's performance parameters, and the dynamic Bradley-Terry model, which also considered the time variable, were applied. Afterwards, We discussed the appropriateness and limitations of the DBT model as a statistical model to examine the relative performance parameters of Korean professional volleyball team for the entire season, along with the trend of performance changes during the season. In conclusion, the DBT model is undoubtedly a very useful analytical method as it not only estimates the performance parameters of each team for the entire season, but also provides accurately inferred performance parameters that reflect the changed performance of the opposing team over time, along with the trend of performance changes. However, the difficulty in analyzing the main factors causing changes in the trend of performance is considered to be a disadvantage of this model.

영어초록

The purpose of this study is to estimate changes in the performance of the Korean professional volleyball team by applying the Dynamic Bradley-Terry model to Korea's volleyball game result records and compare the estimated results with the Static Bradley-Terry model. For the purpose of this study, the match result records of 252 games of 7 professional volleyball teams participating in the 2022-2023 V-League regular season were collected from the website of the Korea Volleyball Federation (KOVO) (https://www.kovo.co.kr/). The data was collected using a web crawling program created by the researcher using Python (python 3.11.2) program. For the collected data, the static Bradley-Terry model, which simply considered the opponent team's performance parameters, and the dynamic Bradley-Terry model, which also considered the time variable, were applied. Afterwards, We discussed the appropriateness and limitations of the DBT model as a statistical model to examine the relative performance parameters of Korean professional volleyball team for the entire season, along with the trend of performance changes during the season. In conclusion, the DBT model is undoubtedly a very useful analytical method as it not only estimates the performance parameters of each team for the entire season, but also provides accurately inferred performance parameters that reflect the changed performance of the opposing team over time, along with the trend of performance changes. However, the difficulty in analyzing the main factors causing changes in the trend of performance is considered to be a disadvantage of this model.

참고자료

· 없음