Population Synthesis with Subregion-Level Control Variable Aggregation

Abstract

This paper details the development of a fully customizable population synthesis program designed to be used for almost any geographic area or input variable. The design addresses several important issues that have been raised regarding other population synthesis methods, including the false-zero cell problem. To address this problem, a routine was developed that allows for the aggregation of control variable categories during execution at the subregional level based on a user-controlled aggregation threshold parameter input, which reduces the areas where false-zero cells are most likely to occur. This procedure also greatly eases the task of data preparation. As part of the study, new validation methods were developed and used to compare the new procedure against other procedures and known test variable distributions. Finally, detailed analysis regarding the trade-off between complexity and computational effort was also undertaken. The use of these new routines and validation methods along with an understanding of the trade-off between complexity and computational effort will allow for the development of more realistic synthetic populations.