Create a Huff model

获得 Business Analyst 许可后可用。

The Huff model is a spatial analysis tool that calculates the potential of a site location using the distance of candidates (customers, clients, patients, or patrons) from the site (such as a store or facility) and its attractiveness. The attractiveness variable is user defined. You can use the Huff model analysis with customer insights to discover why people sometimes travel further for the services they need. You can set the importance or weight of the attractiveness and distance variables. To learn more about the methodology of Huff models, see How Huff Model works. Retail stores, media companies, nonprofit service organizations, and other corporations use the Huff model to determine the probability that a candidate will visit their site.

You can use Huff models to do the following:

• Estimate the market potential.
• Assess the economic impact of a new site location.
• Forecast sales and potential of existing stores or service stations.
• Assess the impact of competitive and environmental changes on outlet performance.

Use the Huff Model tool

The Huff Model tool calculates the probability that a candidate will visit a site based on attractiveness and distance. By default, the Huff model uses exponent values of 1 for attractiveness and -1.5 for distance. The negative exponent values indicate that the further a customer is from the site, the less likely they will go to it. The larger negative exponents cause the probability to drop more steeply as the distance grows.

The default exponent values may not be applicable to your trade or service area. To customize the exponents, use a calibrated Huff model. If distance was not a factor, the distance exponent could be set to 0. A positive distance exponent would indicate that further distance would have a positive impact on customers.

To create a Huff model, do the following:

1. On the Analysis tab, click Business Analysis to open the gallery, and click the Run Huff Model button .

   The Huff Model tool opens in the Geoprocessing pane.

2. Choose the facilities you want to analyze by doing the following:
   1. Use the Input Facility Features drop-down menu to choose the feature layer representing your competitors and existing stores. For this analysis, they are all considered competitions of the new location.
   2. Use the Facility ID Field drop-down menu to choose the unique ID field from the dataset to identify a store or facility location.
3. Choose the candidates you want to analyze by doing the following:
   1. Use the Input Candidate Features drop-down menu to choose a feature layer representing candidate locations. Click the Create button and choose Points if you want to select the new facility point interactively.
   2. Use the Candidate ID Field drop-down menu to choose a unique ID field from the dataset that identifies potential candidates.
4. Choose the sales potential features you want to analyze.

   Sales potential features are the features on which the analysis will be performed. They represent the subdivision of a market into smaller features containing the groups of customers that will be handled as a whole in this model. For example, if your market is a county in the U.S., your sales potential features could be ZIP Codes or census tracts. Sales potential value is a number that represents the maximum amount of sales possible within the given sales potential feature.

   To add the sales potential features, do the following:

   1. Use the Input Sales Potential Features drop-down menu to choose the input point or polygon feature layer used to calculate the sales potential.
   2. Use the Sales Potential ID Field drop-down menu to choose a unique ID field from the dataset to identify sales potential features.
   3. Use the Sales Potential Field drop-down menu to choose the field from the dataset that contains the values that will be used to calculate the sales potential.
5. Use the Output Feature Class parameter to set the name and location of the tool results created.
6. Set the attractiveness variables to define the attractiveness of a site.

   A site may have many attributes that make it attractive to customers. Attractiveness can be computed as a function of the attributes that represent different aspects of attractiveness. It can be expressed as one number that combines all the aspects that make a site attractive.

   1. Set the Existing Facilities Value field to the field that represents the corresponding aspect of attractiveness in the Existing Facilities layer.
   2. Set the Candidates Location Value field to the field that represents the corresponding aspect of attractiveness of a new site.
   3. Set the Exponent field to determine how impactful the variable is to the attractiveness value. The default value is 1.
7. Optionally, enter the following distance parameters:
   • Set Distance Exponent to describe the impact of distance on attractiveness. It is generally a negative number because attractiveness decreases when distance increases. The default value is -1.5.
   • Use the Distance Type drop-down menu to select a distance type.
   • Use the Distance Units drop-down menu to select the distance-measuring units that will be used when calculating distance.
   • If you want to see the what distances are used by the algorithm, use the Output Distance Matrix parameter to set the name and location of the distance calculation matrix table.
8. Optionally, expand the Network Parameters section to set the time of day and time zone of a site location.
9. Click Run .

   A symbolized layer is added to the map. By default, the layer shows areas of high to low probability for candidates to visit a site. You can use the Symbology pane to use predicted sales instead of probability.

Use the Huff Model Calibration tool

The Huff Model Calibration tool helps to calibrate attractiveness and distance exponent values to fit the existing data. This creates more specific and predictive calculations of probability since the default exponent values may not fit your data. Using this tool requires customer data for the existing and competitive locations. If customer data is unavailable, you can use human movement data summarized to your sales potential features.

To calibrate a Huff model, do the following:

1. On the Analysis tab, click Business Analysis to open the gallery, and click the Calibrate Huff Model button .

   The Huff Model Calibration tool opens in the Geoprocessing pane.

2. Choose the facilities you want to analyze by doing the following:
   1. Use the Input Facility Features drop-down menu to choose the feature layer that represents your competitors and existing stores. For this analysis, they are all considered competitions of the new location.
   2. Use the Facility ID Field drop-down menu to choose the unique ID field from the dataset to identify a store or facility location.
3. Choose the customer layer that you want to analyze by doing the following:
   1. Use the Input Customer Features drop-down menu to choose a feature layer that represents customer locations.
   2. Use the Associated Facility ID Field drop-down menu to choose a unique ID field from the customer layer that assigns an individual customer to a facility or store.
   3. Optionally, set the Customer Weight Field for the customer layer. For example, the customer weight could represent the dollar amount that the customer spent at a particular location if it's known.
4. Choose the sales potential features you want to analyze.

   Yo add sales potential features, do the following:

   1. Use the Input Sales Potential Features drop-down menu to choose the input point or polygon feature layer used to calculate the sales potential.
   2. Use the Sales Potential ID Field drop-down menu to choose a unique ID field from the dataset to identify sales potential features.
5. Use the Output Calibration parameter to set the name and location of the output calibration file that will contain the calibrated Huff model results.
6. Set the attractiveness variables to define the attractiveness of the sites.
7. Optionally, enter the following distance parameters:
   • Use the Distance Type drop-down menu to select a distance type.
   • Use the Distance Units drop-down menu to select the distance-measuring units that will be used when calculating distance.
8. Optionally, expand the Network Parameters section to set the time of day and time zone of the site location.
9. Click Run .

   After the tool is successfully executed, the new Huff model item is added to the Catalog pane under Business Analyst. A Huff model item represents the pre-calculated distance and attractiveness exponent values.

To execute a previously calibrated Huff model, do the following:

1. Right-click the existing Huff model item on the Catalog pane to see the context menu.
   1. Click Open to view the Huff Model Calibration dialog box, which displays the detailed information about the Huff model, including its exponents, data source, distance parameters, and errors. Close the Huff Model Calibration dialog box.
   2. Right-click the Huff model item again, then click Run Huff Model to directly feed the calibrated exponents into the Huff model tool. Enter the remaining parameters and click Run .

      The map populates with the symbolized layer representing the Huff model.

2. Optionally, click the Run Calibrated Huff Model button in the Business Analysis gallery for quick access to calibrated Huff models. Choose a previously created calibrated Huff model and run it.

Geoprocessing tools

The Huff model uses the Modeling toolset, including the following tools:

• Huff Model
• Huff Model Calibration

You can use these geoprocessing tools directly in ArcGIS Pro or run them as part of a Python script or a geoprocessing model. Before starting the Huff model tools, if necessary, change the Business Analyst data source.