Analysis of clustering solutions

• US 9,412,109 B2
• Filed: 11/15/2012
• Issued: 08/09/2016
• Est. Priority Date: 11/15/2012
• Status: Active Grant

• Current Assignee: Target Brands Inc. (Target Corporation)
• Original Assignee: Target Brands Inc. (Target Corporation)
• Inventors: Nelson, James Carl, Ranganathan, Raja, Sharma, Abhijit, Sands, Zachary George
• Primary Examiner(s): Gofman, Alex
• Assistant Examiner(s): Gurmu, Muluemebet
• Application Number:
  US13/677,996
• Publication Number:
  US 20140136537A1
• Time in Patent Office: 1,363 Days
• Field of Search: 707/737
• US Class Current: 1/1
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: G06Q 30/00 Commerce

Abstract

A computing system determines incremental values associated with a plurality of clustering solutions. Each of the clustering solutions groups stores of a retailer into clusters in a different way. For each clustering solution in the plurality of clustering solutions, the incremental value associated with the clustering solution indicates a difference between an estimated revenue associated with the clustering solution and revenue associated with a baseline clustering solution. The computing system then determines, based on the incremental values associated with the plurality of clustering solutions, the appropriate number of clusters. The clustering solutions that group the stores into more or fewer clusters than the appropriate number of clusters tend to be associated with incremental values that are the same or lower than the clustering solutions that group the stores into the appropriate number of clusters.

Claims

1. A method comprising:

determining, by a computing system, incremental values associated with a plurality of clustering solutions, wherein each of the clustering solutions groups stores of a retailer into clusters in a different way, wherein for each clustering solution in the plurality of clustering solutions, the incremental value associated with the clustering solution indicates a difference between an estimated revenue associated with the clustering solution and revenue associated with a baseline clustering solution, wherein determining the incremental values associated with the plurality of clustering solutions comprises, for each respective clustering solution from the plurality of clustering solutions;

determining, by the computing system, incremental values associated with each of the clusters into which the respective clustering solution groups the stores, wherein the incremental value associated with each of the clusters into which the respective clustering solution groups the stores is an aggregation of incremental values associated with the stores in the cluster, and the incremental value associated with each store in the cluster is an estimated change in a revenue of the store if a current assortment of items in the store is swapped with another assortment of items; and

determining, by the computing system, the incremental value associated with the respective clustering solution based on a sum of the incremental values associated with each of the clusters into which the respective clustering solution groups the stores;

determining, by the computing system and based on the incremental values associated with the plurality of clustering solutions, an appropriate number of clusters into which to group the stores of the retailer, wherein the appropriate number of clusters corresponds to a turning point of a best fit curve through a plurality of points, each respective point of the plurality of points corresponding to a respective clustering solution of the plurality of clustering solutions, the respective point having a first coordinate corresponding to a number of clusters associated with the respective clustering solution and a second coordinate corresponding to the incremental value associated with the respective clustering solution;

selecting, based at least on part on the appropriate number of clusters, a particular clustering solution in the plurality of clustering solutions; and

distributing, based at least in part on the particular clustering solution, merchandise to the stores.

2. The method of claim 1, wherein determining the incremental values associated with each of the clusters into which the respective clustering solution groups the stores comprises, for each respective cluster into which the respective clustering solution groups the stores:

determining, by the computing system, the incremental values associated with each of the stores in the respective cluster; and

determining, by the computing system, the incremental value associated with the respective cluster based on a sum of the incremental values associated with each of the stores in the respective cluster.

3. The method of claim 2,

wherein a plurality of values of a classification parameter are associated with items sold at the stores in the respective cluster;

wherein determining the incremental values associated with each of the stores in the respective cluster comprises, for each respective store in the respective cluster;

for each value of the classification parameter sold at the respective store;

determining, by the computing system, an actual number of swaps executed for the value of the classification parameter, each of the swaps being a substitution of an assortment of items associated with the value of the classification parameter with an assortment of items associated with a different value of the classification parameter;

determining, by the computing system, an average sales represented by items associated with the value of the classification parameter;

multiplying, by the computing system, the average sales represented by items associated with the value of the classification parameter by the actual number of swaps executed for the value of the classification parameter; and

adding, by the computing system, a product of the multiplication to the incremental value associated with the respective store.

4. The method of claim 3, wherein different values of the classification parameter correspond to different brands.

5. The method of claim 3, wherein determining an actual number of swaps executed for the value of the classification parameter comprises determining, by the computing system and based at least in part on a distance metric for the value of the classification parameter, the actual number of swaps executed for the value of the classification parameter, the distance metric being an indicator of a performance of items associated with the value of the classification parameter.

6. The method of claim 5, further comprising determining, by the computing device, the distance metric as a Cartesian distance from an even productivity line to a point for the value of the classification parameter, a location of the point being based on a first z score and a second z score, the first z score being for a share of spend associated with the value of the classification parameter, the second z score being for a share of items sold at the respective store that are associated with the value of the classification parameter, the even productivity line being a baseline measure of performance that represents a situation where z scores for share of spend is equal to the z scores for share of items.

7. The method of claim 6, further comprising:

determining, by the computing device, the first z score such that the first z score is equal to

8. The method of claim 1, wherein determining the appropriate number of clusters comprises determining, by the computing system, the best fit curve based on a polynomial regression over the plurality of points, wherein the appropriate number of clusters is equal to a number of clusters associated with a maximum point of the curve.

9. A computing system comprising:

one or more processors; and

one or more storage devices that store instructions that, when executed by the one or more processors, cause the computing system to;

determine incremental values associated with a plurality of clustering solutions, wherein each of the clustering solutions groups stores of a retailer into clusters in a different way, wherein for each clustering solution in the plurality of clustering solutions, the incremental value associated with the clustering solution indicates a difference between an estimated revenue associated with the clustering solution and revenue associated with a baseline clustering solution, wherein the instructions cause the computing system to determine the incremental values associated with the plurality of clustering solutions in part by causing the computing system to, for each respective clustering solution from the plurality of clustering solutions;

determine incremental values associated with each of the clusters into which the respective clustering solution groups the stores, wherein the incremental value associated with each of the clusters into which the respective clustering solution groups the stores is an aggregation of incremental values associated with the stores in the cluster, and the incremental value associated with each store in the cluster is an estimated change in a revenue of the store if a current assortment of items in the store is swapped with another assortment of items; and

determine the incremental value associated with the respective clustering solution based on a sum of the incremental values associated with each of the clusters into which the respective clustering solution groups the stores;

determine an optimal cluster count based on the incremental values associated with the plurality of clustering solutions, the optimal cluster count indicating a number of clusters into which the stores are grouped, wherein the optimal cluster count corresponds to a turning point of a best fit curve through a plurality of points, each respective point of the plurality of points corresponding to a respective clustering solution of the plurality of clustering solutions, the respective point having a first coordinate corresponding to a number of clusters associated with the respective clustering solution and a second coordinate corresponding to the incremental value associated with the respective clustering solution; and

select, based at least on part on the appropriate number of clusters, a particular clustering solution in the plurality of clustering solutions, wherein merchandise is distributed to the stores based at least in part on the particular clustering solution.

10. The computing system of claim 9, wherein for each respective cluster into which the respective clustering solution groups the stores, the instructions, when executed by the one or more processors, cause the computing system to:

determine the incremental values associated with each of the stores in the respective cluster; and

determine the incremental value associated with the respective cluster based on a sum of the incremental values associated with each of the stores in the respective cluster.

11. The computing system of claim 10, wherein for each respective store in the respective cluster, the instructions, when executed by the one or more processors, cause the computing system to:

for each brand sold at the respective store;

determine an actual number of swaps executed for the brand;

determine an average sales represented by items associated with the brand;

multiply the average sales represented by items associated with the brand by the actual number of swaps executed for the brand; and

add a product of the multiplication to the incremental value associated with the respective store.

12. The computing system of claim 9, wherein the instructions, when executed by the one or more processors, cause the computing system to determine the best fit curve based on a polynomial regression over the plurality of points, wherein the optimal cluster count is equal to a number of clusters associated with a maximum point of the curve.

13. A non-transitory computer readable storage medium that stores instructions that, when executed by one or more processors of a computing system, cause the computing system to:

determine incremental values associated with a plurality of clustering solutions, wherein each of the clustering solutions groups stores of a retailer into clusters in a different way, wherein for each clustering solution in the plurality of clustering solutions, the incremental value associated with the clustering solution indicates a difference between an estimated revenue associated with the clustering solution and revenue associated with a baseline clustering solution, and as part of causing the computing system to determine the incremental values associated with the plurality of clustering solutions, the instructions cause the computing system to, for each respective clustering solution from the plurality of clustering solutions;

determine incremental values associated with each of the clusters into which the respective clustering solution groups the stores, wherein the incremental value associated with each of the clusters into which the respective clustering solution groups the stores is an aggregation of incremental values associated with the stores in the cluster, and the incremental value associated with each store in the cluster is an estimated change in a revenue of the store if a current assortment of items in the store is swapped with another assortment of items; and

determine the incremental value associated with the respective clustering solution based on a sum of the incremental values associated with each of the clusters into which the respective clustering solution groups the stores;

determine, based on a comparison of incremental values associated with the plurality of clustering solutions, an appropriate number of clusters such that the appropriate number of clusters corresponds to a turning point of a best fit curve through a plurality of points, each respective point of the plurality of points corresponding to a respective clustering solution of the plurality of clustering solutions, the respective point having a first coordinate corresponding to a number of clusters associated with the respective clustering solution and a second coordinate corresponding to the incremental value associated with the respective clustering solution; and

select, based at least on part on the appropriate number of clusters, a particular clustering solution in the plurality of clustering solutions, wherein merchandise is distributed to the stores based at least in part on the particular clustering solution.

14. The non-transitory computer readable storage medium of claim 13, wherein the instructions, when executed by the one or more processors, cause the computing system to determine the best fit curve based on a polynomial regression over the plurality of points, wherein the appropriate number of clusters is equal to a number of clusters associated with a maximum point of the curve.