(eng): We propose how to construct big data-driven macroeconomic fan charts, using machine learning methods to reflect the information in 216 relevant economic variables. Such data-rich fan charts do not rely on restrictive model assumptions and allow the exploration of non-Gaussian, asymmetric, heavy-tailed data and their non-linear interactions. By allowing complex patterns to be learned from a data-rich environment, our fan charts are useful for decision making that depends on the uncertainty of a potentially large number of economic variables — most public policy issues.