Beam-column joints are vital structural elements in buildings and bridges, where accurate prediction of shear force is crucial for ensuring structural integrity and safety. Assessing the shear strength of beam-column joints is complex and requires a combination of theoretical models and analytical methods to account for various structural factors and contexts comprehensively. This paper employs machine learning techniques to predict the shear force in beam-column joints using a comprehensive dataset comprising 473 samples, with 70% allocated for training and 30% for testing. Various algorithms, including regression trees, ensemble boosting, and neural networks, are compared. The performance of these models is evaluated using metrics such as root mean square error (RMSE), mean absolute error (MAE), and mean squared error (MSE). Furthermore, a comparative study was conducted on models trained using various codes of practice, including the American Concrete Institute (ACI), the Indian Standard (IS), and the Architectural Institute of Japan (AIJ). These results provide valuable insights into the effectiveness of machine learning in predicting shear force in beam-column connections, considering different regulatory standards.