Auto clustering prediction models

Patent metadata

Field	Value
Publication number	US-11568432-B2
Application number	US-202016856267-A
Country	US
Kind code	B2
Filing date	Apr 23, 2020
Priority date	Apr 23, 2020
Publication date	Jan 31, 2023
Grant date	Jan 31, 2023

Abstract

Embodiments predict future demand for a first product by receiving historical sales data for an aggregate products/locations level, the historical sales data including a plurality of sales data points, including sales data points for the first product at each of a plurality of locations. Embodiments extract a plurality of different types of features related to sales of each of the products and generate a plurality of clusters of sales data points based on the plurality of different types of features. Embodiments train each of the clusters to generate a plurality of trained cluster models including promotion effects per cluster. For a particular time period, a particular location and the first product, embodiment identify the features for the time period and map to one of the trained cluster models to fetch the promotion effects for the time period. Embodiments then use the promotion effects to forecast demand for the first product.

First claim

What is claimed is:

1. A method of predicting future demand for a first product, the method comprising:
receiving historical sales data for an aggregate products/locations level, the historical sales data comprising a plurality of sales data points, including sales data points for the first product at each of a plurality of locations;
extracting a plurality of different types of features related to sales of each of the products;
generating, using auto clustering, a plurality of clusters of sales data points based on the plurality of different types of features, each of the plurality of clusters comprising a set of the features that are more similar to each other than different set of features corresponding to the other clusters, the auto clustering comprising data mining of the sales data points, wherein each cluster of the plurality of clusters comprises a corresponding mean, and the generating comprises: assigning each of the features to a respective cluster with a nearest mean, recalculating the means for features assigned to each cluster, and repeating the assigning and recalculating until assignments no longer change;
training each of the clusters using the historical sales data to generate a plurality of trained cluster models comprising a set of promotion effects per cluster, the training comprising running a regression for each of the plurality of clusters;
for a particular time period, a particular location and the first product, identifying the features for the time period and mapping to one of the trained cluster models to fetch the corresponding set of promotion effects for the time period; and
using the corresponding set of promotion effects from the mapped trained cluster model to forecast demand for the first product, wherein the mapped trained cluster model predicts demand for the first product;
the method implemented by one or more processors executing instructions.

2. The method of claim 1 , wherein the different types of features comprise one or more of product related features, store features, timing related features, or promotions.

3. The method of claim 1 , wherein the plurality of clusters are generated using k-means clustering, and the assigning an initial set of k means m 1 (1) , . . . , m k (1) comprises:
S i (t) ={x p :|x p −m i (t) ∥ 2 ≤∥x p −m j (t) ∥ 2 ∀j, 1≤ j≤k},

where each x p is assigned to exactly one S (t) , even if it could be assigned to two or more of them.

4. The method of claim 1 , wherein the training each of the clusters comprises running a linear regression on each cluster.

5. The method of claim 1 , further comprising saving each of trained cluster models into a binary file.

6. The method of claim 1 , the using the promotion effect to forecast demand comprising: demand=base demand*promotion lift*price lift.

7. The method of claim 1 , wherein the first product comprises a single stock-keeping unit (SKU) and the historical sales data comprises all SKUs sold at all locations for an entity.

8. The method of claim 1 , further comprising:
in response to the determined forecasted demand for the first product, electronically sending the forecasted demand to an inventory management system;
at the inventory management system, based on the received forecasted demand, generating an electronic order to automatically generate shipments, via a transportation mechanism, of additional quantities of the first product to a plurality of retail stores.

9. A non-transitory computer-readable medium having instructions stored thereon that, when executed by a processor, cause the processor to predict future demand for a first product, the prediction comprising:
receiving historical sales data for an aggregate products/locations level, the historical sales data comprising a plurality of sales data points, including sales data points for the first product at each of a plurality of locations;
extracting a plurality of different types of features related to sales of each of the products;
generating, using auto clustering, a plurality of clusters of sales data points based on the plurality of different types of features, each of the plurality of clusters comprising a set of the features that are more similar to each other than different set of features corresponding to the other clusters, the auto clustering comprising data mining of the sales data points, wherein each cluster of the plurality of clusters comprises a corresponding mean, and the generating comprises: assigning each of the features to a respective cluster with a nearest mean, recalculating the means for features assigned to each cluster, and repeating the assigning and recalculating until assignments no longer change;
training each of the clusters using the historical sales data to generate a plurality of trained cluster models comprising a set of promotion effects per cluster, the training comprising running a regression for each of the plurality of clusters;
for a particular time period, a particular location and the first product, identifying the features for the time period and mapping to one of the trained cluster models to fetch the corresponding set of promotion effects for the time period; and
using the corresponding set of promotion effects from the mapped trained cluster model to forecast demand for the first product, wherein the mapped trained cluster model predicts demand for the first product.

10. The computer-readable medium of claim 9 , wherein the different types of features comprise one or more of product related features, store features, timing related features, or promotions.

11. The computer-readable medium of claim 9 , wherein the plurality of clusters are generated using k-means clustering, and the assigning an initial set of k means m 1 (1) , . . . , m k (1) comprises:
S i (t) ={x p :|x p −m i (t) ∥ 2 ≤∥x p −m j (t) ∥ 2 ∀j, 1≤ j≤k},

where each x p is assigned to exactly one S (t) , even if it could be assigned to two or more of them.

12. The computer-readable medium of claim 9 , wherein the training each of the clusters comprises running a linear regression on each cluster.

13. The computer-readable medium of claim 9 , the predicting further comprising saving each of trained cluster models into a binary file.

14. The computer-readable medium of claim 9 , the using the promotion effect to forecast demand comprising: demand=base demand*promotion lift*price lift.

15. The computer-readable medium of claim 9 , wherein the first product comprises a single stock-keeping unit (SKU) and the historical sales data comprises all SKUs sold at all locations for an entity.

16. The computer-readable medium of claim 9 , the predicting further comprising:
in response to the determined forecasted demand for the first product, electronically sending the forecasted demand to an inventory management system;
at the inventory management system, based on the received forecasted demand, generating an electronic order to automatically generate shipments, via a transportation mechanism, of additional quantities of the first product to a plurality of retail stores.

17. A product demand forecasting system for predicting future demand for a first product comprising:
one or more processors coupled to one or more point of sale systems, the processors receiving historical sales data for an aggregate products/locations level, the historical sales data comprising a plurality of sales data points, including sales data points for the first product at each of a plurality of locations;
the proces

Assignees

• Oracle Int Corp

Inventors

• Lei Ming
• Popescu Catalin

Classifications

• G06K9/6223

  Physics · mapped topic

• G06Q30/0204 Primary

  Market segmentation · CPC title

• G06Q30/0206

  Price or cost determination based on market factors · CPC title

• G06Q10/04 Primary

  Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" (market predictions or forecasting for commercial activities G06Q30/0202) · CPC title

• G06F18/23213

  with fixed number of clusters, e.g. K-means clustering · CPC title