Wuhan Ligong Daxue Xuebao (Jiaotong Kexue Yu Gongcheng Ban)/Journal of Wuhan University of Technology (Transportation Science and Engineering)

Agent-based transportation modeling requires disaggregated data for each individual and household within the study area. It is known that obtaining disaggregated data is very difficult due to privacy and cost concerns. Therefore, it was necessary to create a population synthesis, which is the creation of a synthetic population with attributes that are similar to the attributes of the real population of the study area (Fayoum urban City), based on the population census prepared by the Egyptian Central Agency for Public Mobilization and Statistics (CAPMAS). To address these challenges, population synthesis techniques have been developed to generate privacy-preserving synthetic microdata suitable for microsimulation (i.e., Iterative Proportional Fitting (IPF), Iterative Proportional Updating (IPU), Combinatorial Optimization (CO), and Markov-based and fitness-based syntheses (FBS),). This paper aims to make an efficient population synthesis to generate synthetic microdata for agent-based models, enabling the simulation of travel behavior. This study used the Iterative Proportional Updates (IPU) algorithm, which simultaneously adjusts household- and person-level distributions, building on the traditional Iterative Proportional Fitting (IPF) method. The accuracy of the synthetic populations generated by the IPU depends on the quality of the input data, and new heuristic procedures are introduced for processing Fayoum census data. Encouraging results from using the Iterative Proportional Updating method indicate strong potential to create a more realistic synthetic population with diverse demographic characteristics and to apply it broadly to other geographic areas in Egypt.