Abstract
Synthetic baseline population data is one of the most important data required for the activity based travel demand model. The conventional approach to create this baseline population mainly relies on the Iterative Proportional Fitting (IPF) procedure. However, the traditional IPF procedure assumes the known input data from both the observed cell counts and their marginal counts. This paper presents the application of least square procedure for estimating baseline population distribution in the area where only partial marginal distribution data are available. The method concentrates on optimizing the least squares of the errors between the estimated conditional probability and the target conditional probability, given the constraints of underlying population information in the study area (such as total population, total population by gender, and total population by age etc.). Numerical examples and the case study of Phitsanulok city in Thailand are also presented.
