Working in MXFusion breaks up into two primary phases. Model definition involves defining the variables, distributions, and functions that make up your model. Inference then takes in real values and learns parameters for your model or gives predictions over the data.