Rolling bearings are pivotal components in the realm of rotating machinery, exerting a profound influence on overall system performance. Timely identification of defective bearings is crucial to prevent widespread malfunctions in machinery systems. Rolling element bearings (REBs) constitute integral components in the machinery of diverse industries. To prevent operational breakdowns and damages, it is imperative to establish effective techniques for monitoring the condition and diagnosing faults in these bearings. The advent of machine learning (ML) introduces a novel approach to fault diagnosis in rolling element bearings. In this study, ML models, specifically the K-Nearest Neighbor (K-NN) and Random Forest, are employed to categorize faults in various ball-bearing elements. With accuracy rates of 96% and 97.8%, respectively, these models exhibit robust capabilities in addressing the complexities of fault detection in mechanical systems. The research offers a thorough evaluation of how successfully K-NN and random forests distinguish between distinct fault classes using confusion matrices and graphical representations. Machine learning classifiers are trained to utilize time-domain and frequency-domain features taken from Case Western Reserve University (CWRU) bearing data that is made publicly available. This study not only advances the understanding of fault detection but also serves as a practical guide for implementing machine learning solutions in the realm of mechanical system maintenance.