Search academic texts

Information × Registration Number 2123U006673, Article popup.category Препринт Title popup.author Mamonov Nazarii popup.publication 01-01-2023 popup.source_user Український католицький університет popup.source https://hdl.handle.net/20.500.14570/4813 popup.publisher Description This paper examines the use of machine learning algorithms in forecasting the professional success of new-coming National Basketball Association (NBA) play- ers. The study’s objective was to build a model to forecast NBA success of new- coming players (draftees) and to identify key predictors that significantly influence a player’s NBA trajectory, focusing on variables such as college performance and mock draft position. Utilizing a robust dataset with 655 players drafted from 2000 to 2017, the study employed various machine learning models, including Support Vec- tor Machines (SVM), Random Forests, K-Nearest Neighbors, Logistic Regression, and Extreme Gradient Boosting (XGBoost), each optimized through rigorous hyper- parameter tuning. The results indicate that these machine learning models can predict player suc- cess with considerable accuracy, with the XGBoost model demonstrating superior performance, having accuracy 2 times greater than the accuracy of the benchmark model which was based on current drafting strategies. The analysis also unveiled less obvious factors contributing to a player’s professional trajectory, providing a more nuanced understanding of talent identification in the NBA. The study underscores the potential of machine learning in sports analytics and its transformative implications for strategic decision-making in professional basket- ball. Future research is encouraged to incorporate more diverse data sources and explore emerging machine learning techniques to further enhance the predictive power and insights provided by such models. All code used for this thesis can be found here: Github popup.nrat_date 2025-05-09 Close