Anecdotal and empirical evidence has shown strong associations between the built environment and individuals' travel decision. To date, data about individuals' travel behavior and the nature of the retail environment have not been linked at the fine-grained level for verifying such relationships. GPS and GIS have revolutionized how we measure and monitor land use and individual travel behavior. Compared with traditional travel survey methods, GPS technologies provide more accurate and detailed information about individuals' trips. Based the GPS travel data in the Twin Cities we analyze the impact of individuals' interactions with road network structure and the destinations' accessibility on individuals' destination choice for home-based non-work retail trips. The results reveal that higher accessibility and diversity of services make the destination more attractive. Further, accessibility and diversity of establishments in a walking zone are often highly correlated. A destination reached via a more circuitous or discontinuous route dampens its appeal. In addition, we build an agent-based simulation tool to study retail location choice on a supply chain network consisting of suppliers, retailers, and consumers. The simulation software illustrates that the clustering of retailers can emerge from the balance of distance to suppliers and the distance to consumers. We further applied this tool in the Transportation Geography and Networks course (CE 5180) at the University of Minnesota. Student feedback reveals that it is a useful active learning tool for transportation and urban planning education. The software also has the potential of being extended for an integrated regional transportation-land use forecasting model.