Business Analyst offers two forms of model calibration—using real customer data and using survey data. Calibration of any model is critical to maximize its predictive value. You will need the following to calibrate your model:
- If using real customer data, it must contain information from a sample of households in each subgeography area within the study area. You need customer data for each existing and competitive store location in the study area. The customer information is converted in the model to proportions for each subgeography area.
- Survey data will be used to determine the frequency of shopping trips each respondent makes to the stores within the study area. Exit interviews are effective in collecting this information.
- Ensure that each subgeography area is adequately represented in the sample.
- The Potential Customers geographic level is usually a polygon trade area representing subareas where potential customers live. This can also be a point layer (for example, block centroids) that has associated demographic data.
- The competitive store layer should include all competitive locations in a given study area. This layer should also include any of your existing store locations in the study area, since they will act as competitors to a new store location. In most cases, this layer will be a Business Analyst store layer.
- Competitive store locations can be extracted from the Add Business Listings function in Business Analyst.
- The first step in executing this wizard is to define a study area that includes all the trade areas of all competing stores being analyzed.
- Store attraction fields, also known as predictor values, often include attributes of a store, such as square footage, number of parking spaces, advertising, store hours, prices, age, appearance, signage, and accessibility.
- Click the Business Analyst drop-down menu and click Sales Potential Modeling.
- Click Create Modeling Analysis.
- Click Huff Model Calibration as the type of modeling analysis and click Next.
- Click the first drop-down menu to select the layer containing your sales potential field.
You have the option of checking the Only show Business Analyst data layers (BDS) check box. Business Analyst geography layers contain many fields, including consumer expenditure data, that can be used as indicators of sales potential.
- Click the second drop-down menu and click the ID field for your sales potential layer.
- Click the third drop-down menu and click the sales potential field. Business Analyst geography layers often use a tree structure to present categories of fields that can be expanded to show individual fields in that category.
Hold the Ctrl key and click + or - in the tree to expand or collapse the tree structure.
- Click Next.
- Click the first drop-down menu and click the competitive stores layer.
- Click the second drop-down menu and click the Store ID field.
- Click Next.
- Click one of the methods of selecting a potential store location.
- If you choose By entering address, enter the address of the location by clicking each of the fields in the Value column.
- Click OK to continue.
- If you choose By selecting a feature from layer, click the first drop-down menu and click the layer containing the potential site.
- Click the second drop-down menu and click the individual feature to use for the site location.
- Click Next.
The Calibrate Model utility is covered in detail in the Huff Model Calibration section. It is also provided in the Advanced Huff Model wizard. You can choose the statistically calibrated parameter from a previous analysis or enter the parameters manually.
- Make your choice and click Next.
If you click Enter parameters manually, a dialog box appears to the right.
- Click a method for how distance will be calculated in the model.
Use the + or buttons to add or remove predictor variables.
- After adding a variable, click the variable in the Variable column to activate a drop-down menu for choosing any variable in the layer.
- Click the variable table in the Potential Site Value column and type a value.
- Click the Coefficient column and enter a value between -1.0 and -3.0 to indicate the impact of travel distance on a customer's willingness to travel to make a purchase at the store.
The closer the number is to -1.0, the more willing the customer is to travel a greater distance to make the purchase.
- Click Next.
- Type a name for the new model in the text box, type any comments, then click Finish.
Your new analysis area is created and appears on the map.