Accurate failure time estimation of mechanical components plays a signiﬁcant role in enhancing the reliability of the machines. Ball-bearing failures are the most probable failures in the industrial rotating machinery. Therefore, predicting the remaining useful life of ball bearing proves to be helpful for maintenance scheduling. In data-driven methods for prognostics, the remaining useful lifetime is predicted based on a health indicator. The health indicator detects the condition of equipment or components by monitoring sensor data such as vibration signals. In order to estimate the degradation trends and remaining useful life several machine learning approaches can then be employed for extracting useful features from raw signals.