Qualitative and quantitative comparisons of agent-based and cell-based synthesis estimation methods of base-year data for land-use microsimulations

Abstract

Land-use microsimulation is becoming an indispensable function in a planning support system for sustainable urban development because it provides the detailed information necessary for decision making on emerging issues at the household or firm level. In land-use microsimulations, there are two approaches for estimating base-year micro-data: cell-based population synthesis, which generally uses the iterative proportional fitting method, and agent-based methods. This chapter compares these two methods qualitatively and quantitatively. The qualitative comparison shows that neither one is superior in every aspect. The cellbased method is preferred when the microsimulation deals with data sufficiently simple, while the agent-based method is preferred when accurate and/or numerous micro-data attributes are demanded. Similarly, the quantitative comparison based on a goodness-of-fit evaluation does not show a single superior method for all applications. These findings suggest a way for selecting a better method based on the conditions of the microsimulation model and the purpose of its application.