Advances in population synthesis: fitting many attributes per agent and fitting to household and person margins simultaneously |
| |
Authors: | David R Pritchard Eric J Miller |
| |
Institution: | (1) Metrolinx, 20 Bay St. Suite 901, Toronto, ON, M5S 1A4, Canada;(2) Cities Centre, University of Toronto, 455 Spadina Ave., Suite 400, Toronto, ON, M5S 2G8, Canada |
| |
Abstract: | Agent-based microsimulation models of transportation, land use or other socioeconomic processes require an initial synthetic
population derived from census data, conventionally created using the iterative proportional fitting (IPF) procedure. This
paper introduces a novel computational method that allows the synthesis of many more attributes and finer attribute categories
than previous approaches, both of which are long-standing limitations discussed in the literature. Additionally, a new approach
is used to fit household and person zonal attribute distributions simultaneously. This technique was first adopted to address
limitations specific to Canadian census data, but could also be useful in U.S. and other applications. The results of each
new method are evaluated empirically in terms of goodness-of-fit. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|