Developed a binary classification model to predict machine failures using sensor data (VOC, USS, AQ, etc.). Applied data preprocessing, outlier removal (IQR method), and robust scaling. Achieved 91% accuracy with strong recall for failure cases, enabling insights for preventive maintenance in industrial systems.