排球勝負預測模型之研究—以日本V. League為例

Abstract

排球發展歷史悠久，時至今日仍依靠教練經驗進行球員調度及技戰術調整，因此藉由數據進行分析可使球員調度更加客觀，增加球隊勝率，本研究以日本V1男子組2009-10年至2021-22年共13賽季1440場例行賽為研究對象，蒐集官方提供之各場技術表現數據、選手身體素質數據與主客場為研究樣本，運用邏輯斯迴歸分析探討各項技術表現、選手身體素質與主客場對排球比賽勝負的影響，期望找出能預測排球比賽勝負的關鍵因子，研究結果顯示，以全樣本為例，正確分類率為86.3%，技術表現、身體素質及主客場能有效預測球隊比賽勝負，主場球隊擁有較高獲勝機會，技術表現方面發球得分為影響比賽勝負最關鍵因素，身體素質方面為舉球彈跳高度為最關鍵因素，且在預測模型中，加入球員身體素質之預測模型相較於只放入技術表現之預測模型有較高正確分類率，顯示球員身體素質能有效預測排球比賽之勝負。Volleyball has a long history, player tactical adjustments still rely on the experience of coaches. Therefore, utilizing data analysis can provide a objective approach to player scheduling, ultimately increasing the team's winning percentage. This study focuses on the V1 Men's Division, covering 13 seasons from 2009-10 to 2021-22, with a total of 1440 regular-season matches as the research subjects. The research involves collecting technical data from each match, player physical fitness data, and home/away game information as the study samples. The study employs logistic regression analysis to explore the impact of various performance indicators, player physical fitness, and home/away games on the outcome of volleyball matches. The goal is to identify key factors that can predict the results of volleyball matches. The results of the research indicate that, for the entire sample, the correct classification rate is 86.3%. Technical performance, physical fitness, and home/away games are effective predictors of team match outcomes. Home teams have a higher chance of winning, with serving points being the most crucial factor influencing the match outcome in terms of technical performance. In terms of physical fitness, the jumping height during setter is the most important factor. Moreover, the predict model incorporating player physical fitness demonstrates a higher correct classification rate compared to a model using only technical performance, suggesting that player physical fitness can effectively predict the outcome of volleyball matches in the predict model.