ABSTRACT
In food safety, openness about mistakes is a vital ingredient for high standards. The same concept applies to modelling. Internal validation is the most common form of model evaluation. A common motivation for using spatial microsimulation is lack of data on a specific variable. External validation is more rigorous as it relates simultaneously to the model’s performance and whether the input data are suitable for answering the research questions explored by spatial microsimulation. This chapter explains how to undertake routine checks on spatial microsimulation procedures, how to identify outlying variables and zones which are simply not performing well and how to undertake external validation. J. Pearson’s coeffcient of correlation is the most commonly used measure of aggregate level model fit for internal validation. Root mean squared error is similar to the absolute error metrics, but uses the sum of squares of the error.
