Spatial Measurement of Retail Store Demand

briefly described in the following sections. The general problem of measuring market penetration is then analyzed in light of the comments on these two techniques. Finally, an application of these procedures in the supermarket industry is presented and the results of the application are discussed. TREND SURFACE ANALYSIS Data obtained from areal sampling may be thought to consist of both trend (signal) components and local (noise) components. Trend components refer to the systematic regional patterns of variation while local components consist of nonsystematic chance variations. Attempts to define and isolate the trend component by means of fitting polynomial surfaces to areal data are known as trend surface analysis. When the polynomial surface is computed, the equation may be used to plot an isarithmic map which defines the trend surface. Polynomial surfaces may be constructed in many ways. Two of the more useful approaches are by means of power series polynomials and trigonometric polynomials. If Z is the variable to be estimated and (X, Y) the grid location of the variable, then the power series polynomial is of the form: M N