An application of bootstrapping for determining a decision rule for site location

Abstract

This article provides a general methodology for determining and evaluating a decision rule for hotel site location. Given (1) an indicator of hotel success, (2) an ideal decision rule based on this indicator if it were known without error, and (3) a model for predicting the value of the success indicator at a proposed site, we propose a procedure for finding the optimal model-based decision rule for any specified optimality criterion and for evaluating the worth of the rule. The methodology is based on the statistical technique called bootstrapping. This method reduces the bias of conventional methods of estimation that have been applied in the context of site-location modeling. The methodology is illustrated using data from a large hotel chain in the United States and evaluated using an independent evaluation sample.