The prioritization of green and sustainable transportation is growing globally. Shared bicycle systems effectively contribute to this goal and serve as a valuable connection to other transportation modes, particularly in urban areas. This study employs the XGBoost machine learning method to predict bike rental demand using seven input variables. The XGBoost model is enhanced by the MSA algorithm to optimize runtime and performance. The model’s predictive power is assessed through various statistical metrics, including R2, RMSE, MAE, and MAPE. Results indicate that the XGBoost model demonstrates strong predictive performance and accuracy in predicting bike rental demand, with R2 = 0.878, RMSE = 63.474, MAE = 42.596, and MAPE = 0.505. Furthermore, the analysis of SHAP values reveals the travel patterns and preferences of public bike system users. To facilitate the use of this machine learning model, a graphical user interface (GUI) has been developed.