Methods for handling missing values abound, but they are almost all focused in applications in which the goal is effect estimation, e.g. calculation and statistical inference for regression coefficients. Our focus here is instead on prediction.
Note that this should not be confused with packages that use regression methods for inputation. There, the missing values in a variable Xi may be inputed using predictions in which X2 is regressed on the other Xj. Instead, we are interested in applications in prediction itself is the focus, such as forecasting or disease diagnosis.