To leverage the available system data, factories across the globe are embracing Predictive Maintenance in the era of Industry 4.0. This has put much emphasis on the development of data-driven algorithms to enable company-wide deployment of predictive maintenance.  In the current study, the objective is to predict failures by mining asset maintenance log. Towards this end, first, a text mining-based approach has been suggested to identify different failures, followed by a survival analysis based approach to predict the failure sequence and patterns. The developed algorithm has been tested on maintenance logs of centrifugal casting machines.