The continuing evolution of Industry 4.0 is bringing new opportunities for companies to create value. The goal of this paper is to identify a data-driven approach for the predictive maintenance of Resistance Spot Welding (RSW), a common technique for joining sheet metals, through the real time monitoring of the electrode wear state, which is one of the most important parameters that influences the quality of a weld. The data used for the analysis was collected from sensors mounted on an RSW machine in the J-Tech laboratory at Politecnico di Torino, and then several features were extracted from it. Python and its libraries were used to perform machine learning and deep learning techniques.