RETAIL SALES OVERLAPPING PROMOTIONS FORECASTING USING AN OPTIMIZED P-NORM

• US 20150120381A1
• Filed: 10/24/2013
• Published: 04/30/2015
• Est. Priority Date: 10/24/2013
• Status: Active Application

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: POPESCU, Catalin, HE, Lin, LEI, Ming, TANG, Chun
• Application Number:
  US14/062,140
• Publication Number:
  US 20150120381A1
• Time in Patent Office: Days
• US Class Current: 705/7.31
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities

Abstract

A system that generates a sales forecast for an item receives the sales history for prior sales periods that includes at least one stand-alone time period when a single promotion event is active, and at least one overlapping time period when two or more promotion events are active and overlapping. For each stand-alone time period, the system determines a stand-alone lift for each promotion event active during the stand-alone time periods. For each overlapping time period, the system determines a combined overlap lift of promotion events that are overlapping using a p-norm.

Claims

1. A computer-readable medium having instructions stored thereon that, when executed by a processor, cause the processor to generate a sales forecast for an item, the generating the sales forecast comprising:

receiving sales history for prior sales periods comprising at least one stand-alone time period when a single promotion event is active, and at least one overlapping time period when two or more promotion events are active and overlapping;

for each stand-alone time period, determining a stand-alone lift for each promotion event active during the stand-alone time periods; and

for each overlapping time period, determining a combined overlap lift of the promotion events that are overlapping using a p-norm.

2. The computer-readable medium of claim 1, the generating the sales forecast further comprising:

generating a baseline sales forecast;

generating a promotional sales forecast comprising combining the baseline sales forecast, the stand-alone lifts and the overlap lifts.

3. The computer-readable medium of claim 1, the generating the sales forecast further comprising:

generating a baseline sales forecast;

generating a promotional sales forecast comprising multiplying the baseline sales forecast, the stand-alone lifts and the overlap lifts.

4. The computer-readable medium of claim 1, wherein the determining the combined overlap lift comprises receiving a value of the p-norm.

5. The computer-readable medium of claim 1, wherein the determining the combined overlap lift comprises optimizing a value of the p-norm.

6. The computer-readable medium of claim 5, wherein the optimizing comprises finding the value of the p-norm that minimizes the Mean Absolute Percentage Error.

7. The computer-readable medium of claim 6, wherein the optimizing comprises for overlapping promotions X₁ with effect β

1 and X₂ with effect β

2;

8. The computer-readable medium of claim 6, wherein the optimizing comprises for overlapping promotions X₁ with effect β

1 and X₂ with effect β

2;

9. A method of generating a sales forecast for an item, the method comprising:

receiving sales history for prior sales periods comprising at least one stand-alone time period when a single promotion event is active, and at least one overlapping time period when two or more promotion events are active and overlapping;

for each stand-alone time period, determining a stand-alone lift for each promotion event active during the stand-alone time periods; and

for each overlapping time period, determining a combined overlap lift of the promotion events that are overlapping using a p-norm.

10. The method of claim 9, further comprising:

generating a baseline sales forecast;

generating a promotional sales forecast comprising combining the baseline sales forecast, the stand-alone lifts and the overlap lifts.

11. The method of claim 9, further comprising:

generating a baseline sales forecast;

generating a promotional sales forecast comprising multiplying the baseline sales forecast, the stand-alone lifts and the overlap lifts.

12. The method of claim 9, wherein the determining the combined overlap lift comprises receiving a value of the p-norm.

13. The method of claim 9, wherein the determining the combined overlap lift comprises optimizing a value of the p-norm.

14. The method of claim 13, wherein the optimizing comprises finding the value of the p-norm that minimizes the Mean Absolute Percentage Error.

15. The method of claim 14, wherein the optimizing comprises for overlapping promotions X₁ with effect β

1 and X₂ with effect β

2;

16. The method of claim 14, wherein the optimizing comprises for overlapping promotions X₁ with effect β

1 and X₂ with effect β

2;

17. A sales forecasting system comprising:

an overlap determination module that receives sales history for prior sales periods comprising at least one stand-alone time period when a single promotion event is active, and at least one overlapping time period when two or more promotion events are active and overlapping; and

a promotion lift generator that for each stand-alone time period, determines a stand-alone lift for each promotion event active during the stand-alone time periods, and for each overlapping time period, determines a combined overlap lift of the promotion events that are overlapping using a p-norm.

18. The sales forecasting system of claim 17, further comprising:

a forecast module that generates a baseline sales forecast and generates a promotional sales forecast comprising combining the baseline sales forecast, the stand-alone lifts and the overlap lifts.

19. The sales forecasting system of claim 17, further comprising:

a forecast module that generates a baseline sales forecast and generates a promotional sales forecast comprising multiplying the baseline sales forecast, the stand-alone lifts and the overlap lifts.

20. The sales forecasting system of claim 17, wherein the determine the combined overlap lift comprises optimizing a value of the p-norm.