The banking industry is experiencing rapid growth, particularly in telemarketing strategies to increase product and service sales. Despite widespread use, these strategies need higher success rates due to data imbalance, where fewer customers accept offers than those who reject them. This study evaluates machine learning algorithms, including Random Forest, Gradient Boosting, Extra Trees, and AdaBoost, without and handling imbalanced data using the Random Over-Sampling Examples (ROSE) method. The evaluation covers accuracy, precision, recall, F1-score, and AUC of the ROC curve. Results indicate that Random Forest and AdaBoost consistently perform well, with Random Forest maintaining a high accuracy of 91.00% after handling imbalanced data. Gradient Boosting and Extra Trees improve in precision post-oversampling. All models exhibit high AUC values, close to 0.94, demonstrating excellent differentiation between positive and negative classes. The study concludes that addressing data imbalance enhances model performance, making these models suitable for effective telemarketing strategies in the banking sector.