A Multi-Scale Suitability Analysis of Home-Improvement Retail-Store Site Selection for Ontario, Canada

Abstract

A multi-scale suitability analysis using big data (4.7 million suitability scores) is presented across a large spatial extent (1.076 million km2) to identify potential locations for new home-improvement retail stores. Suitability scores were generated for individual property parcels using criteria weights derived from surveyed retail-industry experts. To increase capacity for site selection, distributions of suitability scores were generated at census dissemination areas (populations 500-700; n = 19,963) and census metropolitan and agglomeration areas (core populations >10,000; n = 43). Analogues among metropolitan and agglomeration areas were generated and spatial clustering was used to identify groups of highly-suitable parcels within urban areas. Lastly, individual parcels can be interrogated for overall suitability or individual criteria scores. Our approach combines retail methods typically used in isolation (e.g. location quotient, Huff’s model, network analysis) and demonstrates how a simple survey can be used to weight criteria. Results show that survey respondents were in general agreement and that top-line revenues were more critical to perceived location success than development and operational costs. Analysis of suitability scores found analogues and clusters of census metropolitan areas that coincide with store sales as well as clusters of highly suitable parcels predominantly located around major highways. In addition to identifying challenges and solutions to the presented research, we also describe future research directions that extend the presented static analysis to include processes like store closure and openings, competition, and land use change through the use of agent-based modelling.