Basketball Shot Analysis Based on Goal Assembly Disturbance

Basketball player’s performance and skills can be enhanced with an intelligent feedback system that uses machine learning and sensor data. One of the sources for this kind of system is IMU (Inertial measurement unit) sensor data that is obtained from the basketball goal assembly which comprises of the backboard, rim, and netting. In this paper, the correlation of successful free throw shots with the nature of the physical disturbance the shot causes to the basketball goal assembly is investigated. A machine learning-based model that predicts the success of the shots made was developed and proposed for implementation, using data collected from the goal assembly. For feature selection, machine learning model training, and accuracy comparison, the MATLAB 2021 classification learner is employed. In addition, using various machine learning models, a classification model for shot success based on disturbance of possible combinations of the three parts of the goal assembly is investigated. According to a preliminary experiment findings and examinations of the data gathered, the SVM classifier can accurately distinguish made or missed free throws with an accuracy of 87.7%. Furthermore, the ensemble tree classifier predicted missed and successful shots, with detail of where the ball touched before passing through the net(pure net; rim and net; all), with an accuracy of 76.1%. These classification models can be used as a baseline to improve players’ shooting performance because our experiments demonstrate a strong correlation between shot success and sensor data. It can also be used as a source of additional information for the growth of modern basketball training and coaching