Shared Bicycle Systems (SBC) have contributed to the improvement of urban mobility and have become increasingly common in cities. However, for the proper functioning of these systems, it is necessary to analyze the users’ habits so that it is possible to supply the demand for stations with high demand and avoid wasting resources in underused stations, collecting bicycles in overcrowded stations and redistributing them in stations that are low on stock or even empty. The rebalancing bicycles process (Bike Sharing Rebalancing Problem - BSRP) requires efficient redistribution strategies, which depends on the modeling and forecasting of rental demand, evidencing the relevance of approaches that assist in decision-making and strategic management of the companies responsible for these systems. In this work, a data set from Capital BikeShare (Washington, D.C.) is analyzed, exploring the relationship of meteorological and seasonal conditions in this service. Regression and classification models were developed using classic literature algorithms as Linear Regression (LR), Support Vector Machine (SVM), Gaussian Naive Bayes (NB), K-Nearest Neighbors (KNN), MultiLayer Perceptron (MLP), Random Forest (RF) and Decision Tree (DT). The objective is to predict the number of rentals, providing a parameter to assist in the efficient redistribution of bicycles in SBC. The use of multiple algorithms has also been tried to build homogeneous and heterogeneous committees. The performances of the generated predictive models were analyzed using 10-part cross-validation. Friedman and Nemenyi statistical tests also were applied. Moreove, descriptive models using k-means to group rental stations were also discussed, highlighting such groupings as a useful technique in improving the performance of predictive models. The best regression model was generated with RF and has determination coefficient of 0.963, while the best classification model was developed generated with DT and has an accuracy of 96.21%. An application scenario for potential predictive models has been described to facilitate the interpretation of the prediction results for the managers of these services and to improve the decision-making process in SBC