New Methodology for Synthesizing Population in Metropolitans

Abstract

In the Activity Based Modeling (ABM) approach, an activity pattern is specifically assigned to each individual in a household. In this way, one of the fundamental steps in the ABM approach is to project the socio-economic characteristics of all household members while considering some marginal constraints available for the whole of population which is known as population synthesizing. In the current paper, the household characteristics distribution is defined as the probability of having a household with a particular size, number of students and workers. The main goal of current research is to statistically fit the household characteristics distribution of the population to a previously obtained distribution for sample households in a way which satisfies the marginal constraints in traffic zones. It is also followed to solve two main issues in most previous research on population synthesis, one of them is related to the so called "zero cell" and the other is known as the "Integrality" problem. Similarity between characteristics distributions of the sample and all households can be achieved by using Maximum Likelihood Estimation (MLE) with the marginal constraints. Satisfying all marginal constraints in a single optimization for a real case study involving a huge number of households increases the mathematical complexity of the problem, and likely leads to an infeasible state. In the current paper, a new idea for solving this problem in real cases is proposed. The proposed algorithm using GAMS software is implemented in Mashhad city (population of more than 2.5 million) in Iran.