An Investigation of the Transferability of Trip Generation Models and the Utilization of aSpatial Context Variable

The cost of collecting data for travel demand modeling is high and increasing each year. Data collection costs could easily exceed the annual budget of a metropolitan planning organization (MPO) in small or medium-sized area. Many of these agencies borrow or transfer data and/or models from other areas since they cannot afford the cost of collecting local data. This study included two primary research objectives. The first was to test the appropriateness of transferring commonly used trip generation models from one urban area to another under specific circumstances. The second was to improve the transferability of models by including a variable reflecting the spatial context of households, the basic unit of trip generation used in most MPO models. The data utilized for this research were drawn from four separate travel surveys and included data for 11 metropolitan planning areas in two states.

The key finding of this research was that a meaningful consistent measure of spatial context can be included in trip generation models, and it can make these models more generic and transferable. This finding that the transferability of trip production models can be improved by including an additional variable called “Area Type” should be helpful to many MPOs, which have to borrow models or survey data from other areas. The data needed for developing this variable should not pose any difficulty since it is based on population data, which is readily available from the US Census Bureau. Further, the algorithm needed to stratify grid cells and the households located in them into different categories of Area Type is available in many geographic information system software packages including TransCAD, which is widely used by MPOs of all sizes.