The current study aims to assess One-Class Support Vector Machines’ (One-Class SVM) effectiveness in detecting anomalies in datasets pertaining to motor vibrations. A number of datasets including motor vibration data under various circumstances—both healthy and defective—must be loaded and pre-processed before the procedure can begin. Each of the datasets basically consists of X, Y, and Z axes of vibration data taken for a particular period of time. A key aspect of our work is the application of Grid Search to determine the optimal hyperparameters for the One-Class SVM, including the nu, kernel, and gamma parameters. For each dataset, the best model is selected based on a cross-validated F1-score. In this way we create numerous One-Class SVM models based on each dataset and validate them on the test set using some evaluation parameters such as accuracy, precision, recall, and F1-score for each dataset. The dataset achieving the highest F1-score is identified as the best-matching dataset for the test set used. By understanding the conditions and fault scenarios of the best-performing dataset, we can infer the specific faults that are most effectively detected by the model in the test data. Our findings provide insights into the effectiveness of One-Class SVMs for detecting anomalies in motor vibration data, demonstrating the importance of careful parameter tuning and the potential for this method to identify fault types in an induction motor.