ABSTRACT
This chapter briefly describes other techniques for spatial microsimulation. IPF is just one strategy for obtaining a spatial microdataset. However, researchers tend to select one method that they are comfortable with and stick with that for their models. An interesting alternative to IPF method is the GREGWT algorithm. First implemented in the SAS language by the Statistical Service area of the Australian Bureau of Statistics (ABS), the algorithm reweighs a set of initial weights using a Generalized Regression Weighting procedure (hence the name GREGWT). In general terms, an optimization problem consists of a function, the result of which must be minimised or maximised, called an objective function. This function is not necessarily defined for the entire domain of possible inputs. The base R function optim provides a general purpose optimization framework for numerically solving objective functions. Different reweighting strategies are suitable in different contexts and there is no clear winner for every occasion.
