Smart Manufacturing (SM) can be defined as the extensive application of computer-integrated manufacturing and advanced intelligence systems to enable rapid manufacturing of new products, dynamic response to product demand, and real-time optimization of manufacturing production and supply-chain networks. For this reason, SM is now attracting a huge interest in both academic and industrial communities and will probably drive the manufacturing evolution in the next decade. In SM, data play a key role. They can support decisional systems and human operators by helping them to improve production and process control, to monitor continuous production flows, to prevent or detect equipment failures at an early stage, to minimize inefficiencies through the overall supply chain, and so on. In fact, data can be exploited by combining a wide variety of advanced technologies to give machines the ability to learn, adapt, make decisions, and display new behaviours. In this regard, the aim of the study concerns the proposal of a data-driven framework to predict the electrode wear in Resistance Spot Welding process. Electrode wear is the most important factor that introduce high variability and uncertainty in the quality of spot welds. Using an equipped medium-frequency welding machine, various data such as thermal maps of the spot surfaces by passive thermography, electrode surface diameters, electrodes-workpiece contact conditions, process variables, and electrode displacement curves can be collected. These data can be provided as input to a Machine Learning algorithm to predict electrode wear over time, thus ensuring a reliable spot weld process and joint quality.