Publications-Theses
Article View/Open
Publication Export
-
題名 職業網球單打評分模型的實證研究
An Empirical Study of Rating-System Model on Professional Tennis作者 蕭立承
Hsiao, Li-Chen貢獻者 余清祥<br>洪英超
Yue, Ching-Syang<br>Hung, Ying-Chao
蕭立承
Hsiao, Li-Chen關鍵詞 運動大數據
探索性資料分析
評分模型
貝氏分析
職業網球
Sport Big Data
Exploratory Data Analysis
Rating Model
Bayesian Analysis
Professional Tennis Matches日期 2020 上傳時間 2-Sep-2020 11:43:15 (UTC+8) 摘要 預測是決策分析的重要課題，如果能夠清楚地掌握未知狀況，減少因應意外事件所需的心力與資源，則更能有效率地解決問題。預測對於職業運動及球類格外重要，經常用於設計訓練課程、安排隊形及對戰策略，可以提升個人表現及增加獲勝的機會，現在國內外有不少博弈業者也以預測為研究議題，根據球隊及球員戰績及相關資料評估勝率，採用統計或機器學習模型計算賠率。本文以預測男女職業網球大滿貫（四大公開賽：澳洲、法國、溫布敦、美國）的勝負為目標，透過探索性資料分析（Exploratory Data Analysis）尋找較為重要的解釋變數，比較統計學習及機器學習等量化模型的成效。另外，本文也引進職業西洋棋常用的Glicko模型，研擬改進這個模型的可能性；其中，Glicko評分模型由哈佛教授Mark Glickman提出，依據貝氏理論更新球員特性。本文先透過探索性資料分析，尋找較能反映比賽勝負的球員相關變數，以此作為建立統計及機器學習的基礎，之後再將最佳模型與Glicko模型比較。本文採用2000～2019年男女職業網球四大滿貫資料，採用分類模型如羅吉士迴歸（統計學習模型）、SVM、Neural Network及Lightgbm（以上三者為機器學習模型），透過交叉驗證評估優劣。分析發現職業網球排名與比賽勝負關係最為密切，單以此變數訓練模型準確性可達7成，而Glicko模型在準確性或AUC（Area Under Curve）都有不錯的表現，用於男性或女性的勝負預測都優於統計及機器學習模型。本文嘗試進一步優化Glicko模型，綜合各場地類別的Glicko及其他解釋變數，發現可略微增加Glicko模型的預測準確性。
Prediction is important in decision analysis and the problem solving would be more efficient if we can narrow the possibilities down. Prediction is also important in professional sports. It can be used in designing training courses, arranging gaming strategies, and organizing team members, in order to improve game performance and winning probability. Many bookmakers use statistical or machine learning models to predict the winning odds, based on match records and related data. In this study, our goal is to investigate the models of predicting the match outcomes of Grand Slam tournaments (Australian Open, French Open, Wimbledon Championships, and US. Open). In particular, we will apply Exploratory Data Analysis (EDA) to explore important variables. In addition to statistical and machine learning models, we also consider Glicko rating model, commonly used in professional chess, to predict the game results. Glicko was proposed by Harvard professor Mark Glickman and it updates player rating based on Bayesian theory.The empirical study is based on men’s and women’s Grand Slam data (2000~ 2019). We first use EDA to determine important variables and then apply classification models, such as logistic regression (statistical learning model), Support Vector Machine, Neural Network and Light Gradient Boosting Machine (machine learning model), to evaluate the classification results through cross-validation. Our analysis results show that the professional tennis ranking is the most important variable and all models include this variable can achieve at least 70% of accuracy. The Glicko model outperforms statistical and machine learning models, with respect to accuracy and AUC (Area Under Curve). However, the improvement of modified Glicko model is quite limited.參考文獻 英文文獻1.Barnett, T. and Clarke, S. R. (2005). Combining Player Statistics to Predict Outcomes of Tennis Matches. IMA Journal of Management Mathematics, 16(2):113-120.2.Boulier, B. L. and Stekler, H. O. (1999). Are Sports Seedings Good Predictors ? : An Evaluation. International Journal of Forecasting, 15(1):83-91.3.Bradley, R. A. and Terry, M. E. (1952). The rank analysis of incomplete block designs: 1, The method of paired comparisons. Biometrika, 39, 324-345.4.Cornman, A., Spellman, G. and Wright, D. (2017). Machine Learning for Professional Tennis Match Prediction and Betting.5.Elo, A. E. (1978). The Rating of Chess players, Past and Present. New York: Arco.6.Herbrich, R., Minka, T. and Graepel, T. (2006). Trueskill(tm): A Bayesian Skill Rating System. In Advances in Neural Information Processing Systems, pp. 569-576.7.Huang, T. K., WENG, R. C. and LIN, C.J. (2006). Generalized Bradley-Terry models and multi-class probability estimates. J. Mach. Learn. 85-115.8.Glickman, M. E. (1999). Parameter Estimation in Large Dynamic Paired Comparison Experiments. Applied Statistics, 48(3):377-394.9.Gilsdorf, K. F. and Sukhatme, V. A. (2008). Testing rosen`s sequential elimination in tournamento model incentives and player performance in professional tennis. Journal of Sports Economics, 9:287-303.10.Kovalchik, S. A. (2016). Searching for the goat of tennis win prediction. Journal of Quantitative Analysis in Sports, 12:127-138.11.Klaassen, F. and Magnus, J. (2003). Forecasting the winner of a tennis match. European Journal of Operational Research, 148:257-267.12.Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q. and Liu, T. Y. (2017). Lightgbm: A highly efficient gradient boosting decision tree. In Advances in Neural Information Processing Systems, 3149-3157.13.Lisi, F., and Zanella, G. (2017). Tennis Betting: Can Statistics Beat Bookmakers ? Electronic Journal of Applied Statistical Analysis, 10:790–808.14.Martin, I. (2019). A Point-based Bayesian Hierarchical Model to Predict the Outcome of Tennis Matches. Journal of Quantitative Analysis in Sports, 313-325.15.Newton, P. K. and Keller, J. B. (2005). Probability of Winning at Tennis I. Theory and Data. Studies in Applied Mathematics, 114(3):241-269.16.Pollard, G.N., Cross, R., and Meyer, D. (2006). An analysis of ten years of the four Grand Slam men’s singles data for lack of independence of set outcomes. Journal of Sports Science and Medicine, 5, 561-566.17.Rosenblatt, F. (1958). The Perceptron: A Probabilistic Model For Information Storage And Organization In The Brain. Psychological Review, 65 (6): 386-408.18.Srivastava, S. (2019). Predicting success probability in professional tennis tournaments using a logistic regression model. Advances in Analytics and Applications, 59–65.19.Sipko, M. and Knottenbelt, W. (2015). Machine learning for the prediction of professional tennis matches. 描述 碩士
國立政治大學
統計學系
107354024資料來源 http://thesis.lib.nccu.edu.tw/record/#G0107354024 資料類型 thesis dc.contributor.advisor 余清祥<br>洪英超 zh_TW dc.contributor.advisor Yue, Ching-Syang<br>Hung, Ying-Chao en_US dc.contributor.author (Authors) 蕭立承 zh_TW dc.contributor.author (Authors) Hsiao, Li-Chen en_US dc.creator (作者) 蕭立承 zh_TW dc.creator (作者) Hsiao, Li-Chen en_US dc.date (日期) 2020 en_US dc.date.accessioned 2-Sep-2020 11:43:15 (UTC+8) - dc.date.available 2-Sep-2020 11:43:15 (UTC+8) - dc.date.issued (上傳時間) 2-Sep-2020 11:43:15 (UTC+8) - dc.identifier (Other Identifiers) G0107354024 en_US dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/131478 - dc.description (描述) 碩士 zh_TW dc.description (描述) 國立政治大學 zh_TW dc.description (描述) 統計學系 zh_TW dc.description (描述) 107354024 zh_TW dc.description.abstract (摘要) 預測是決策分析的重要課題，如果能夠清楚地掌握未知狀況，減少因應意外事件所需的心力與資源，則更能有效率地解決問題。預測對於職業運動及球類格外重要，經常用於設計訓練課程、安排隊形及對戰策略，可以提升個人表現及增加獲勝的機會，現在國內外有不少博弈業者也以預測為研究議題，根據球隊及球員戰績及相關資料評估勝率，採用統計或機器學習模型計算賠率。本文以預測男女職業網球大滿貫（四大公開賽：澳洲、法國、溫布敦、美國）的勝負為目標，透過探索性資料分析（Exploratory Data Analysis）尋找較為重要的解釋變數，比較統計學習及機器學習等量化模型的成效。另外，本文也引進職業西洋棋常用的Glicko模型，研擬改進這個模型的可能性；其中，Glicko評分模型由哈佛教授Mark Glickman提出，依據貝氏理論更新球員特性。本文先透過探索性資料分析，尋找較能反映比賽勝負的球員相關變數，以此作為建立統計及機器學習的基礎，之後再將最佳模型與Glicko模型比較。本文採用2000～2019年男女職業網球四大滿貫資料，採用分類模型如羅吉士迴歸（統計學習模型）、SVM、Neural Network及Lightgbm（以上三者為機器學習模型），透過交叉驗證評估優劣。分析發現職業網球排名與比賽勝負關係最為密切，單以此變數訓練模型準確性可達7成，而Glicko模型在準確性或AUC（Area Under Curve）都有不錯的表現，用於男性或女性的勝負預測都優於統計及機器學習模型。本文嘗試進一步優化Glicko模型，綜合各場地類別的Glicko及其他解釋變數，發現可略微增加Glicko模型的預測準確性。 zh_TW dc.description.abstract (摘要) Prediction is important in decision analysis and the problem solving would be more efficient if we can narrow the possibilities down. Prediction is also important in professional sports. It can be used in designing training courses, arranging gaming strategies, and organizing team members, in order to improve game performance and winning probability. Many bookmakers use statistical or machine learning models to predict the winning odds, based on match records and related data. In this study, our goal is to investigate the models of predicting the match outcomes of Grand Slam tournaments (Australian Open, French Open, Wimbledon Championships, and US. Open). In particular, we will apply Exploratory Data Analysis (EDA) to explore important variables. In addition to statistical and machine learning models, we also consider Glicko rating model, commonly used in professional chess, to predict the game results. Glicko was proposed by Harvard professor Mark Glickman and it updates player rating based on Bayesian theory.The empirical study is based on men’s and women’s Grand Slam data (2000~ 2019). We first use EDA to determine important variables and then apply classification models, such as logistic regression (statistical learning model), Support Vector Machine, Neural Network and Light Gradient Boosting Machine (machine learning model), to evaluate the classification results through cross-validation. Our analysis results show that the professional tennis ranking is the most important variable and all models include this variable can achieve at least 70% of accuracy. The Glicko model outperforms statistical and machine learning models, with respect to accuracy and AUC (Area Under Curve). However, the improvement of modified Glicko model is quite limited. en_US dc.description.tableofcontents 第一章 緒論 1第一節 研究動機 1第二節 研究目的 3第二章 文獻探討與研究方法 5第一節 文獻回顧 5第二節 資料介紹 6第三節 研究方法 9第三章 統計學習與機器學習模型 14第一節 探索性資料分析 15第二節 模型建構 26第三節 模型評估 32第四章 Glicko 評分模型 36第一節 Glicko 評分分析 37第二節 集成法（Ensemble Methods） 40第三節 混合評分模型（Mixture-Rating Model） 42第五章 結論與建議 48第一節 結論 48第二節 建議 49參考文獻 51附錄 53Glicko演算法推導 53 zh_TW dc.format.extent 2754194 bytes - dc.format.mimetype application/pdf - dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0107354024 en_US dc.subject (關鍵詞) 運動大數據 zh_TW dc.subject (關鍵詞) 探索性資料分析 zh_TW dc.subject (關鍵詞) 評分模型 zh_TW dc.subject (關鍵詞) 貝氏分析 zh_TW dc.subject (關鍵詞) 職業網球 zh_TW dc.subject (關鍵詞) Sport Big Data en_US dc.subject (關鍵詞) Exploratory Data Analysis en_US dc.subject (關鍵詞) Rating Model en_US dc.subject (關鍵詞) Bayesian Analysis en_US dc.subject (關鍵詞) Professional Tennis Matches en_US dc.title (題名) 職業網球單打評分模型的實證研究 zh_TW dc.title (題名) An Empirical Study of Rating-System Model on Professional Tennis en_US dc.type (資料類型) thesis en_US dc.relation.reference (參考文獻) 英文文獻1.Barnett, T. and Clarke, S. R. (2005). Combining Player Statistics to Predict Outcomes of Tennis Matches. IMA Journal of Management Mathematics, 16(2):113-120.2.Boulier, B. L. and Stekler, H. O. (1999). Are Sports Seedings Good Predictors ? : An Evaluation. International Journal of Forecasting, 15(1):83-91.3.Bradley, R. A. and Terry, M. E. (1952). The rank analysis of incomplete block designs: 1, The method of paired comparisons. Biometrika, 39, 324-345.4.Cornman, A., Spellman, G. and Wright, D. (2017). Machine Learning for Professional Tennis Match Prediction and Betting.5.Elo, A. E. (1978). The Rating of Chess players, Past and Present. New York: Arco.6.Herbrich, R., Minka, T. and Graepel, T. (2006). Trueskill(tm): A Bayesian Skill Rating System. In Advances in Neural Information Processing Systems, pp. 569-576.7.Huang, T. K., WENG, R. C. and LIN, C.J. (2006). Generalized Bradley-Terry models and multi-class probability estimates. J. Mach. Learn. 85-115.8.Glickman, M. E. (1999). Parameter Estimation in Large Dynamic Paired Comparison Experiments. Applied Statistics, 48(3):377-394.9.Gilsdorf, K. F. and Sukhatme, V. A. (2008). Testing rosen`s sequential elimination in tournamento model incentives and player performance in professional tennis. Journal of Sports Economics, 9:287-303.10.Kovalchik, S. A. (2016). Searching for the goat of tennis win prediction. Journal of Quantitative Analysis in Sports, 12:127-138.11.Klaassen, F. and Magnus, J. (2003). Forecasting the winner of a tennis match. European Journal of Operational Research, 148:257-267.12.Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q. and Liu, T. Y. (2017). Lightgbm: A highly efficient gradient boosting decision tree. In Advances in Neural Information Processing Systems, 3149-3157.13.Lisi, F., and Zanella, G. (2017). Tennis Betting: Can Statistics Beat Bookmakers ? Electronic Journal of Applied Statistical Analysis, 10:790–808.14.Martin, I. (2019). A Point-based Bayesian Hierarchical Model to Predict the Outcome of Tennis Matches. Journal of Quantitative Analysis in Sports, 313-325.15.Newton, P. K. and Keller, J. B. (2005). Probability of Winning at Tennis I. Theory and Data. Studies in Applied Mathematics, 114(3):241-269.16.Pollard, G.N., Cross, R., and Meyer, D. (2006). An analysis of ten years of the four Grand Slam men’s singles data for lack of independence of set outcomes. Journal of Sports Science and Medicine, 5, 561-566.17.Rosenblatt, F. (1958). The Perceptron: A Probabilistic Model For Information Storage And Organization In The Brain. Psychological Review, 65 (6): 386-408.18.Srivastava, S. (2019). Predicting success probability in professional tennis tournaments using a logistic regression model. Advances in Analytics and Applications, 59–65.19.Sipko, M. and Knottenbelt, W. (2015). Machine learning for the prediction of professional tennis matches. zh_TW dc.identifier.doi (DOI) 10.6814/NCCU202001670 en_US
