Create Territories from Optimal Locations

Create territories from optimal locations allows you to create territories without seed points. This command divides geographic objects of the alignment layer into clusters with equal weight and selects the best location of each cluster as the territory center. Clusters can be calculated using spatial locations of features of the alignment layer only or in combination with an attribute.

The spatial locations are the centers of geographic objects, for example, centers of ZIP Code polygons. In areas with high concentrations of geographic objects, the clusters have smaller geographic sizes, and more clusters can be created in these areas. Most sources of real data have higher concentrations in cities or populated places, which can be used to create better-balanced territories. Cities will have a number of smaller territories that compensate for larger populations or income in cities. The centroid of clusters calculated by spatial locations is the optimal territory center.

If choosing to locate territory centers using spatial locations only on a constant grid, the territory centers will be located in an even pattern. This does not take into consideration the realistic locations, such as heavily populated areas or the most diverse sections of an area.

Result of an example for using locate territory centers using spatial locations only

Using this same concept on actual source boundaries, the spatial locations create clusters based on the varying sizes. The higher-concentrated areas are given the most weight. The new territory centers are placed at these optimal locations.

Result of an example using locate territory centers using actual source boundaries

If a summary attribute is added to the calculation of clusters, the algorithm will use a modified distance to calculate concentrations. The cluster element with the largest value of selected summary attribute will be used as the optimal territory center.

The image below is an example of cells of equal size; the darker cells represent larger populations. Because cells have equal sizes, the attribute weight is the key parameter. Clustering by spatial locations and the summary attribute of total population creates clusters that are distributed over the most populated places.

The result of an example of cells of equal size

Clustering by spatial locations and the summary attribute of total population creates clusters and centers in areas with high concentrations of the ZIP Codes. These are more pronounced in concentrated areas with large populations.

The result of an example of clustering by spatial locations

Differences of Create territories from optimal locations and Create territories from centers of density

As an example, suppose you are looking to expand your stores into Pennsylvania in the most populated places. The goal is to find the best locations for the stores. The resulting territories can be used later for analyzing these locations. The image below uses the Create territories from optimal locations option.

The result of using Create territories from optimal locations

Some of the resulting territories are located in ideal places, but some territories are located far from the urban areas, which is not ideal because of the distance from other stores. While the coverage of the entire state is good, the more concentrated store network is desired, because all features within Pennsylvania were used in the calculation, and the located centers were pulled away from the cities by the rural features. The image below uses the Create territories from centers of density option.

The result using Create territories from centers of density

This result is better, because territory centers are located in the higher-concentrated areas. Although stores are not distributed across the entire state, the most opportunity is in the urban areas.

Optimal by Capacity

Formula:

Algorithm

where N - optimal number of territories, Sum k - summary value of kth variable at specified or max extent, Capacityk - value of capacity of kth variable, Tolerancek - value of tolerance of kth variable (from 0 to 0,99).

Algorithm:

  1. Calculate summary value for each variable in specified or max area extent.
  2. Find optimal number of territories for each variable (sum/variable value).
  3. Calculate weight of each variable based on tolerance value (tolerance shows importance of variable, less tolerance - more important variable).
  4. Calculate average optimal number using variables weight.
3/7/2014