Abstract:
|
Human migration patterns present a rich set of modeling challenges, including very large data sets and many covariates that may directly or indirectly influence migration. With the advent of the American Community Survey, we are provided with finer spatial detail and shorter time intervals than were previously available in U.S. population migration data as well as social, economic, and geographical variables that may inform migration. Furthermore, migration can be stratified by gender, race, or socioeconomic status. We develop a flexible multivariate spatio-temporal model for migration that incorporates covariates and their uncertainty. The model is designed for multivariate human migration, but is capable of modeling ecological migration, as well as non-standard "migration" problems such as traffic flow, and trade flow.
|