Estimating small area demand for online package delivery

Abstract Using publicly available microdata sets, we show how estimates for online delivery purchases can be generated for small geographic areas defined in our study as micro analysis zones (MAZ) and how these estimates vary across the MAZs that featured in our study. With a focus on Miami-Dade County, we use both the national household travel survey (NHTS) data and synthetic data obtained from Southeast Florida Regional Planning Model (SERPM) to generate demand estimates of online delivery purchases for more than 5300 distinct geographic units in Miami-Dade County. We assess the quality of the estimates using measures of predictive accuracy and by comparing the cumulative values obtained with the population estimates generated from the NHTS survey data for Miami-Dade County. Our approach fills a void in the area of purchases of online delivery items where rich observable data are typically unavailable and it also provides the added potential benefit of being easily replicated nationwide given the emphasis on the use of publicly available data.