Gross margin recovery with supervised machine learning technique

Patent metadata

Field	Value
Publication number	US-11599895-B2
Application number	US-201916514259-A
Country	US
Kind code	B2
Filing date	Jul 17, 2019
Priority date	Jul 17, 2019
Publication date	Mar 7, 2023
Grant date	Mar 7, 2023

Abstract

A mechanism is provided to proactively forecast gross margin for a business unit of an organization utilizing machine learning techniques. Embodiments provide a cascading-architecture machine-learning model to predict gross margin for a period (e.g., an upcoming quarter), utilizing metrics both internal and external to the organization. Internal metrics can include list price change, discounting change, cost impact, and the like. External metrics can include customer information such as propensity to purchase and purchase consumption.

First claim

What is claimed is:

1. A computer-implementable method for generating a predicted gross margin for an organization, the method comprising:
generating an internal factors gross margin prediction using one or more product attributes;
generating an external factors gross margin prediction using one or more customer attributes and customer sales pipeline information;
training a plurality of demand forecast models using a training dataset, the training dataset comprising product demand data across a plurality of periods of time, the product demand data across the plurality of periods of time allowing the plurality of demand forecast models to be trained to generate respective forecasted demand;
selecting a demand forecast model from the plurality of demand forecast models based upon the respective forecasted demand generated using the product demand dataset, the demand forecast model providing a best mean average percent error (MAPE); and,
generating the predicted gross margin for the organization using the demand forecast model, the demand forecast model the internal factors gross margin prediction and the external factors gross margin prediction;
and wherein,
the generating the internal factors is performed by an internal gross margin processor of a gross margin system, generating the external factors is performed by an external gross margin processor of the gross margin system and generating the predicted gross margins is performed by a gross margin processor of the gross margin system, the gross margin system is specifically configured to perform a gross margin predictions operation; and,
the generating the internal factors, generating the external factors and generating the predicted gross margins are performed via a supervised machine learning operation, the supervised machine learning operation using internal parameters and external parameters when generating the predicted gross margin for the organization using the demand forecast model.

2. The method of claim 1 wherein the one or more product attributes comprise one or more of average list price per unit, average cost per unit, average discount per unit, run rate revenue, and routes to market revenue.

3. The method of claim 1 wherein generating the internal factors gross margin prediction further comprises:
forecasting demand for one or more products using one or more forecast models.

4. The method of claim 3 wherein said generating the internal factors gross margin prediction further comprises using a linear regression model to combine information from the one or more forecast models and the one or more product attributes.

5. The method of claim 3 , wherein the one or more forecast models comprise one or more of an autoregressive integrated moving average model, a triple exponential smoothing model, and an additive model comprising a saturation growth model and a piecewise linear model.

6. The method of claim 5 wherein said using the one or more forecast models comprises selecting a best fit forecast model for recent demand data.

7. The method of claim 1 wherein the one or more customer attributes comprise one or more of purchase frequency associated with a customer account, last purchase time associated with the customer account, buying power associated with the customer account, and share of wallet associated with the customer account.

8. The method of claim 1 wherein said generating the external factors gross margin prediction using one or more customer attributes and customer sales pipeline information further comprises:
generating a propensity score for a selected customer account;
calculating a customer account gross margin for a set period if the propensity score is greater than a gross margin threshold;
adjusting the propensity score in light of the customer sales pipeline information associated with the selected customer account, if the propensity score is less than the gross margin threshold; and
calculating the customer account gross margin for the set period if the adjusted propensity score is greater than the gross margin threshold.

9. The method of claim 8 wherein said generating the propensity score comprises using a logistic regression model on the customer attributes for the selected customer account.

10. The method of claim 9 wherein said adjusting the propensity score comprises:
rejecting the customer account for gross margin determination if one of no pipeline exists for the customer account or a pipeline probability associated with the pipeline sales information is less than a first threshold; and
adjusting the propensity score by a first amount if the pipeline sales information is greater than the first threshold.

11. The method of claim 9 wherein said adjusting the propensity score further comprises:
adjusting the propensity score by a second amount if the pipeline sales information is greater than a second threshold, wherein the second threshold is greater than the first threshold and the second amount is greater than the first amount; and
adjusting the propensity score by a third amount if the pipeline sales information is greater than a third threshold, wherein the third threshold is greater than the second threshold and the third amount is greater than the second amount.

12. A specialized computing device comprising:
a processor;
a data bus coupled to the processor; and
a non-transitory, computer-readable storage medium coupled to the data bus, and storing computer program code, wherein the computer program code interacts with a plurality of computer operations and comprises instructions executable by the processor such that the processor is configured to:
generate an internal factors gross margin prediction using one or more product attributes;
generate an external factors gross margin prediction using one or more customer attributes and customer sales pipeline information;
train a plurality of demand forecast models using a training dataset, the training dataset comprising product demand data across a plurality of periods of time, the product demand data across the plurality of periods of time allowing the plurality of demand forecast models to be trained to generate respective forecasted demand;
select a demand forecast model from the plurality of demand forecast models based upon the respective forecasted demand generated using the product demand dataset, the demand forecast model providing a best mean average percent error (MAPE); and,
generate the predicted gross margin for the organization using the demand forecast model, the demand forecast model the internal factors gross margin prediction and the external factors gross margin prediction; and wherein
the generating the internal factors is performed by an internal gross margin processor of a gross margin system, generating the external factors is performed by an external gross margin processor of the gross margin system and generating the predicted gross margins is performed by a gross margin processor of the gross margin system, the gross margin system is specifically configured to perform a gross margin predictions operation; and,
the generating the internal factors, generating the external factors and generating the predicted gross margins are performed via a supervised machine learning operation, the supervised machine learning operation using internal parameters and external parameters when generating the predicted gross margin for the organization using the demand forecast model.

13. The specialized computing device of claim 12 , wherein the processor is configured to generate the internal factors gross margin prediction by being further configured to forecast demand for one

Assignees

• Dell Products Lp

Inventors

• Sinha Moloy Kumar
• Dan Rajkumar

Classifications

• 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

• G06Q30/0202 Primary

  Market predictions or forecasting for commercial activities · CPC title

• G06N20/00

  Machine learning · CPC title