This project focuses on the development of a data-driven methodology that aims to gain more knowledge about welding processes performed on car bodies. This methodology is fundamental to the transition from a periodic maintenance scheme to a predictive maintenance one, which would allow the optimisation of the maintenance scheme and the increase of money and energy savings and of the quality of the production. Moreover, the methodology draws inspiration from Data Mining concept, in order to better match with the technologies and techniques that are being developed in the context of industry 4.0.