Xi Zheng
ABSTRACT
In recent years, extensive analysis has been applied to predicting NBA game results due to the popularity of basketball and massive financial transactions in NBA betting. The primary objective of this research is to construct a predictive model to precisely forecast the outcome of NBA basketball games in the latest 2020-21 and 2021-22 NBA regular season. We designed features which incorporates both external and internal factors such as teams’ Elo rating, average team performance in recent games, home court advantage and tiredness due to back-to-back games. We built up three feature sets and performed feature selection using sequential feature selection (SFS) and recursive feature elimination (RFE) to verify their effectiveness. Results show that novel features such as level of tiredness and difference of Elo ratings between home and away team improves prediction accuracy. To make fair prediction for the latest NBA season, we utilize 10-fold cross validation to train and select models with decent mean accuracy and low standard deviation for final evaluation. It is found that our best random forest model performs fairly well in predicting games in the latest 2020-21 and 2021-22 season with an accuracy of 67.98%. The prediction results and the identification of key features that exert the most significant effects on the results can be helpful and meaningful to different stakeholders in this field, such as team coaches, players and NBA betters.
- Cao C (2012) Sports data mining technology used in basketball outcome prediction. Dublin Institute of Technology.Google Scholar
- Thabtah, F., Zhang, L., & Abdelhamid, N. (2019). NBA game result prediction using feature analysis and machine learning. Annals of Data Science, 6(1), 103-116.Google Scholar
- Nguyen, N. H., Nguyen, D. T. A., Ma, B., & Hu, J. (2021). The application of machine learning and deep learning in sport: predicting NBA players’ performance and popularity. Journal of Information and Telecommunication, 1-19.Google Scholar
- Hvattum, L. M., & Arntzen, H. (2010). Using ELO ratings for match result prediction in association football. International Journal of forecasting, 26(3), 460-470.Google Scholar
- https://projects.fivethirtyeight.com/2018-nba-predictions/games/Google Scholar
- http://www.nbaminer.com/Google Scholar
- Ferrario, A. BASKETBALL ANALYTICS: THE USE OF DATA SCIENCE TO DESCRIBE AND PREDICT THE PERFORMANCE OF AN NBA TEAM.Google Scholar
- Chen, W. J., Jhou, M. J., Lee, T. S., & Lu, C. J. (2021). Hybrid Basketball Game Outcome Prediction Model by Integrating Data Mining Methods for the National Basketball Association. Entropy, 23(4),477Google Scholar
- Loeffelholz, B., Bednar, E., & Bauer, K. W. (2009). Predicting NBA games using neural networks. Journal of Quantitative Analysis in Sports, 5(1).Google Scholar
- Pai, P. F., ChangLiao, L. H., & Lin, K. P. (2017). Analyzing basketball games by a support vector machines with decision tree model. Neural Computing and Applications, 28(12), 4159-4167.Google Scholar
Recommendations
Prediction for NCAA championship
ICBDR '21: Proceedings of the 5th International Conference on Big Data ResearchIn recent years, predicting sports results has become increasingly popular. Basketball is one of the most popular sports in the world, especially in USA, the NCAA game attract lot of fans around over 1000 colleges. In this paper, a new machine learning ...
A Neural Network Model of the NBA Most Valued Player Selection Prediction
PRAI '19: Proceedings of the 2019 the International Conference on Pattern Recognition and Artificial IntelligenceThis study analyzed all the performance of the players in the National Basketball Association (NBA) during a particular season and predicted the most valued players (MVP) of that season. The NBA game is the most popular basketball game all over the ...
Comments