Anthony Spagnolo

The purpose of this study is to illustrate the value of applying spatial analytical techniques to key business issues faced by retail firms. Particularly, the approach to be evaluated and discussed throughout this paper will provide a retail firm with a practical method of integrating the valuable information collected on a daily basis with readily available spatial data to form a decision-support system that can increase the operational efficiency of a retail network and drive effective decision-making. The analysis investigates the use of a multi-method approach to segment a store network into operational store typologies and estimate store performance using a major Canadian specialty retailer as a case study. Specifically, k-means cluster analysis is used to define the portfolio of locations into homogeneous groups of stores using a combination of site-specific, market-specific, and competitive measures while linear multivariate regression analysis is conducted to produce sales forecasting models. The overall goal of the research study is to derive store segments to be used as sub-samples for regression modeling where the store cluster models will be compared to a global model to identify whether the process of clustering a store network will produce more robust, parsimonious, and meaningful models thereby increasing the effectiveness of decision-support activities.