Enhancing the prediction of shot success in NBA Basketball games using machine learning techniques - FNN neural network

Abstract

The advent of data-driven decision-making has sparked a transformation in the sports industry, where the precision of predictive models now serves as a pivotal factor in both team success and financial viability. This thesis examines Machine Learning and Deep Learning models for predicting NBA shot success, with team members developing Random Forest, XGBoost, Feedforward and Recurrent Neural Network models. Notably, the Recurrent Neural Network, previously unapplied in this context, emerged with superior predictive accuracy. This study's primary contribution is unveiling the RNN's potential for shot prediction, paving the way for its future integration into sports strategic planning and business analytics.