Making Household Microsimulation of Travel and Activities Accessible to Planners

Abstract

There is a large gap between the aggregate, trip-based models used by transportation planning agencies and the activity-based, microsimulation methods being espoused by those at the forefront of research. The modeling environment presented here aims to bridge this gap by providing a palatable way for planning agencies to move towards advanced methods. Three components to bridging the gap are emphasized, including an incremental approach, demonstration of clear gains, and providing an environment that eases initial implementation and allows for expansion. The modeling environment (called STEP2) is a household microsimulator developed in TransCAD, which can be used to implement a 4-step model as well as models with longer term behavior and trip chaining. An implementation for Southern Nevada is described and comparisons are made with the region's aggregate 4-step model. The models perform similarly on numerous counts. A key advantage to the microsimulator is that it provides impacts by socioeconomic group (essential for equity analysis) and individual trip movements (for use in a vehicle microsimulator). A sensitivity analysis indicates that the microsimulation model has less inelastic cross elasticity of transit demand with respect to auto travel times than the aggregate model (aggregation error). The trade-off is that microsimulators have simulation error; results are presented regarding the severity of this error. This work shows that shifting to microsimulation does not necessarily require substantial investment to achieve many of the benefits. One of the greatest advantages is a flexible environment that can expand to include additional sensitivity to demographics and transportation policy variables.

From Aggregate Methods to Microsimulation

“…Walker (2005) capably illustrated how the two approaches' computation time and calibration efforts were quite similar, yet the microsimulation model offered the additional benefits of preserving demographic distinctions across the population, thus allowing for analysis of subgroup impacts (of transportation policies and investments, for example). In addition, the microsimulation approach eliminated aggregation errors (spatial and demographic), while allowing for calculations of the associated simulation errors/variability (Walker, 2005). The current paper expands this comparison to a microscopic model with tours, rather than trips, as the basic unit of analysis, and utilizing more sophisticated methods to incorporate intra-household constraints and activity scheduling.…”

“…Walker's (2005) models of Las Vegas, Nevada offered a direct comparison, though the microsimulation model relied on a tripbased format (which does not allow for mode, scheduling or intra-household consistency). Walker (2005) capably illustrated how the two approaches' computation time and calibration efforts were quite similar, yet the microsimulation model offered the additional benefits of preserving demographic distinctions across the population, thus allowing for analysis of subgroup impacts (of transportation policies and investments, for example). In addition, the microsimulation approach eliminated aggregation errors (spatial and demographic), while allowing for calculations of the associated simulation errors/variability (Walker, 2005).…”

Mode Choice Modeling Using Personalized Travel Time and Cost Data

“…Traditionally transportation related models were designed to deal with aggregate information or specific market segments; however, with advancement in technology, much more information has become available to researchers as the demand for more policy sensitive models has increased. These advances have made microsimulation frameworks very popular in the past two decades (1)(2)(3)(4)(5)(6).…”

Population Synthesis: Comparing the Major Techniques Using a Small, Complete Population of Firms

“…Technical aspects such as the ordering of conditional probabilities in the selection of attributes are explored fully, and an example of a synthetic population of households is presented. Variations of the IPFSR technique, generally making use of the iterative proportional fitting (IPF) technique for the creation of multiway tables, are widespread (see, e.g., Smith et al 1995; Beckman et al 1996; Huang and Williamson 2002; Frick and Axhausen 2004; Walker 2004; Ballas et al 2005; Simpson and Tranmer 2005; Arentze, Timmermans, and Hofman 2007; Guo and Bhat 2007). It is important to note that, while IPFSR makes use of IPF, the two techniques are not equivalent.…”