The Third IAEA Technical Meeting on Fusion Data Processing, Validation and Analysis took place at the IAEA Headquarters in Vienna, Austria, from 28–31 May and brought together more than 60 scientists and engineers from 19 Member States, the European Commission and ITER Organisation working on data analysis and machine learning (ML) methods for the processing of fusion data, collected either from experimental diagnostics or from plasma simulations.'Accurate data processing leads to a better understanding of the physics related to nuclear fusion research. It is essential for the careful estimate of the error bars of the raw measurements and processed data,' said D. Mazon (Chair, CEA, France) in his introductory talk and progress has been shown in this direction during the meeting. In particular the meeting discussed new developments in fusion R&D applications in the following areas: inversion techniques, such as tomography; magnetic topology reconstruction, such as equilibrium reconstructions; system identification; scaling laws determination and their accuracy for extrapolation from current machines to fusion reactors; model-based algorithms for control applications; identification of spurious and undesired events, such as disruptive phenomena or hot spots in infra-red images using sophisticated mathematical techniques like neural network and support vector machines.Discussions also focused on the potential use of these techniques for ITER, in particular, on their relevance to the first ITER plasma. The use of Integrated Modelling & Analysis Suite (IMAS) infrastructure—a convenient platform that allows implementation of different simulation codes in the same format—and synthetic diagnostics that could be coupled with this structure, in order to test the different mathematical approaches developed by the meeting participants, were explored in detail. Recent results and progress in the development of new tools will be discussed during the next meeting on the topic, scheduled in 2021.