機器學習於籃球員得分預測之應用與分析

The Application of Machine Learning Approaches for Forecasting Points of Basketball Players

戴聖紘(Sheng-Hong Dai)

指導教授 ： 白炳豐

國立暨南國際大學/管理學院/資訊管理學系/碩士(2020年)

https://doi.org/10.6837/ncnu202000137

摘要

國際的體育競技以及休閒活動中，籃球屬於世界範疇內之前幾名，甚至各國家都為籃球這項運動建立專屬的協會以及聯盟在國內外進行推廣演變至今成為了人類運動史上不可或缺的球類運動之一。由於NBA在國際聲譽上的影響，吸引了來自世界各地的數百萬球迷觀看NBA近乎30個左右的團隊在常規的82場比賽互相競爭，最後於季後賽與冠軍賽決出冠軍隊伍。本研究的目的就是透過根據球員的數據並使用機器學習的方法去預測出NBA聯賽球員之得分預測，得分可以說是左右球場的關鍵要素，分數越高的球隊即是勝利隊伍，那麼找出可以為球團獲取分數的球員就顯得至關重要，而得分數越多的球員也將被視為明星球員，也會是各球團所爭相競奪的重點之一。本研究基於此目的，使用近年來盛行的機器學習方法極限梯度提升(XGBoost)對近年NBA之數據進行球員得分預測，且該方法於財經、醫療、商業以及生產領域皆有廣泛的應用，而本研究使用XGBoost方法進行NBA球員得分預測，也會與其它不同的機器學習算法與XGBoost相互比較，可得出NBA球員得分預測之最佳模型並提供球團對於球員之比較分析以及未來球團的球員管理決策之考量。

關鍵字

籃球 ； 機器學習 ； 國家籃球協會 ； 極端梯度提升 ； 隨機森林 ； 倒傳遞神經網路 ； 廣義回歸神經網路

並列摘要

In the international sports and leisure activities, basketball belongs to the top few in the world. and even countries have established exclusive associations and leagues for basketball, which has become one of the indispensable ball games in the history of human sports. Because of the reputation of NBA in the world, NBA has attracted people from all over the world watch nearly 30 teams compete with each other in 82 regular games, and finally decide the champion team in the playoffs and championships. The purpose of this study is to predict the scores of NBA players by using the data of players and the method of machine learning. The scores can be said to be the key elements of the left and right game. The teams with higher scores are It's a winning team, so it's very important to find out the players who can get scores for the team, and the players who get more scores will also be regarded as star players, which will also be one of the key points for each team to compete for. Based on this purpose, this study uses the popular machine learning method, XGBoost to predict the scores of NBA players in recent years. And this method is widely used in finance, medical, commercial and production fields. XGBoost is used to predict NBA players' scores, which is also different from other machine learning algorithms, By comparing with each other, we can get the best model of NBA players' score prediction, provide the team for the comparative analysis of players and the future team player management decision-making considerations.

並列關鍵字

Basketball ； Machine learning ； NBA ； XGBOOST ； RF ； BPNN ； GRNN

參考文獻

參考文獻

1. Agrawal, S., Singh, S. P., & Sharma, J. K. (2018). Predicting Results of Indian Premier League T-20 Matches using Machine Learning. Paper presented at the 2018 8th International Conference on Communication Systems and Network Technologies (CSNT).

2. Baboota, R., & Kaur, H. (2019). Predictive analysis and modelling football results using machine learning approach for English Premier League. International Journal of Forecasting, 35(2), 741-755.

3. Chakraborty, D., & Elzarka, H. (2019). Early detection of faults in HVAC systems using an XGBoost model with a dynamic threshold. Energy and Buildings, 185, 326-344.

4. Chao, L., Wen-hui, Z., & Ji-ming, L. (2019). Study of Star/Galaxy Classification Based on the XGBoost Algorithm. Chinese Astronomy and Astrophysics, 43(4), 539-548.