Broad-crested rectangular gabion weirs are used as an effective means of regulating flow to an impoundment while conserving large amounts of energy and preventing adverse ecological impacts in hydraulic structures. This study aimed to develop various machine learning algorithms to estimate the coefficients of discharge for free-flowing water over the through-weir. Many of the machine learning regression algorithms were created from experimental data: RF, SR, SVR, XGBoost, and CatBoost, to find the best fit to model the coefficients of discharge from the dataset, with CatBoost and XGBoost providing the best overall performance. The total dataset was split so that 20% was used to test how well the models were performing, and the remaining 80% was used to train the models to provide the most accurate measure of performance for the models. The XGBoost model provided the best performance with an R2 of 0.96, NSE of 0.95, and RMSE of 0.029. The performance of the Simple Regression and Support Vector Regression models was less than desirable for each condition observed in comparison to that of the XGBoost model’s performance. Finally, the 80:20 ratio of training-testing data sets was shown to be a large contributor to the predictive accuracy of the empirical relationship of the final equation produced by the GEP models.