System and method for item-level demand forecasts using linear mixed-effects models

Patent metadata

Field	Value
Publication number	US-2016239776-A1
Application number	US-201514622252-A
Country	US
Kind code	A1
Filing date	Feb 13, 2015
Priority date	Feb 13, 2015
Publication date	Aug 18, 2016
Grant date	—

Abstract

A system and method for forecasting sales is presented. A method might begin by receiving a request to produce a demand forecast for a stock keeping unit (SKU). Then, the SKU is placed in one or more clusters. A cluster seasonality profile is calculated for each of the one or more clusters. An item seasonality profile is calculated for the SKU. Then the demand forecast for the SKU is generated. The demand forecast is adjusted using the cluster seasonality profile for each of the one or more clusters and the item seasonality profile for the SKU. Then inventory can be ordered based on the adjusted demand forecast. Other embodiments are also disclosed herein.

First claim

What is claimed is:

1 . A method comprising:
receiving a request to produce a demand forecast for a stock keeping unit (SKU); placing the SKU in one or more clusters; calculating a cluster seasonality profile for each of the one or more clusters; calculating an item seasonality profile for the SKU; generating the demand forecast for the SKU; adjusting the demand forecast using the cluster seasonality profile for each of the one or more clusters and the item seasonality profile for the SKU; and ordering inventory based on the adjusted demand forecast.

2 . The method of claim 1 wherein:
adjusting the demand forecast using the cluster seasonality profile comprises:
using a mixed effect model to calculate an adjustment to the demand forecast, the mixed-effect model using the following equation, where Y is a matrix for the adjustment to the demand forecast, X is a known design matrix relating β to Y, Z is a known design matrix relating Y to μ, β is a coefficient representing a seasonality of the cluster, μ is a coefficient representing a seasonality of the SKU, and ε is an error term:
Y=Xβ+Zμ+ε.

3 . The method of claim 2 wherein:
the β coefficient represents a weighting for sales data for the cluster over a one week period of time; and
the μ coefficient represents a weighting for sales data for the SKU over a one week period of time.

4 . The method of claim 2 wherein:
X is a matrix of two different clustering methods;
β is a coefficient representing a seasonality of two or more clusters, calculated using the following equation, where β 1 is the seasonality of a first cluster and β 2 is the seasonality of a second cluster:
β=(β 1 β 2 ) T .

5 . The method of claim 1 wherein:
the one or more clusters comprise at least a category-based cluster and a semantic-based cluster.

6 . The method of claim 1 wherein:
generating the demand forecast for the SKU comprises using a moving average technique to estimate the demand forecast.

7 . The method of claim 1 wherein:
calculating the cluster seasonality profile for each of the one or more clusters comprises:
using a mixed-effect model to remove randomness of cluster data to calculate the cluster seasonality profile.

8 . A system comprising:
a user input device; a display device; one or more processing modules; and one or more non-transitory storage modules storing computing instructions configured to run on the one or more processing modules and perform the acts of:
receiving a request to produce a demand forecast for a stock keeping unit (SKU);
placing the SKU in one or more clusters;
calculating a cluster seasonality profile for each of the one or more clusters;
calculating an item seasonality profile for the SKU;
generating the demand forecast for the SKU;
adjusting the demand forecast using the cluster seasonality profile for each of the one or more clusters and the item seasonality profile for the SKU; and
ordering inventory based on the adjusted demand forecast.

9 . The system of claim 8 wherein:
adjusting the demand forecast using the cluster seasonality profile comprises:
using a mixed effect model to calculate an adjustment to the demand forecast, the mixed-effect model using the following equation, where Y is a matrix for the adjustment to the demand forecast, X is a known design matrix relating β to Y, Z is a known design matrix relating Y to μ, β is a coefficient representing a seasonality of the cluster, μ is a coefficient representing a seasonality of the SKU, and ε is an error term:
Y=Xβ+Zμ+ε.

10 . The system of claim 9 wherein:
the β coefficient represents a weighting for sales data for the cluster over a one week period of time; and
the μ coefficient represents a weighting for sales data for the SKU over a one week period of time.

11 . The system of claim 9 wherein:
X is a matrix of two different clustering methods;
β is a coefficient representing a seasonality of two or more clusters, calculated using the
following equation, where β 1 is the seasonality of a first cluster and β 2 is the seasonality of a second cluster:
β=(β 1 β 2 ) T .

12 . The system of claim 8 wherein:
the one or more clusters comprise at least a category-based cluster and a semantic-based cluster.

13 . The system of claim 8 wherein:
generating the demand forecast for the SKU comprises using a moving average technique to estimate the demand forecast.

14 . The system of claim 8 wherein:
calculating the cluster seasonality profile for each of the one or more clusters comprises:
using a mixed-effect model to remove randomness of cluster data to calculate the cluster seasonality profile.

15 . At least one non-transitory memory storage module having computer instructions stored thereon executable by one or more processing modules to:
receive a request to produce a demand forecast for a stock keeping unit (SKU); place the SKU in one or more clusters; calculate a cluster seasonality profile for each of the one or more clusters; calculate an item seasonality profile for the SKU; generate the demand forecast for the SKU; adjust the demand forecast using the cluster seasonality profile for each of the one or more clusters and the item seasonality profile for the SKU; and order inventory based on the adjusted demand forecast.

16 . The at least one non-transitory memory storage module of claim 15 wherein:
adjusting the demand forecast using the cluster seasonality profile comprises:
using a mixed effect model to calculate an adjustment to the demand forecast, the mixed-effect model using the following equation, where Y is a matrix for the adjustment to the demand forecast, X is a known design matrix relating β to Y, Z is a known design matrix relating Y to μ, β is a coefficient representing a seasonality of the cluster, μ is a coefficient representing a seasonality of the SKU, and ε is an error term:
Y=Xβ+Zμ+ε.

17 . The at least one non-transitory memory storage module of claim 15 wherein:
the β coefficient represents a weighting for sales data for the cluster over a one week period of time; and
the μ coefficient represents a weighting for sales data for the SKU over a one week period of time.

18 . The at least one non-transitory memory storage module of claim 16 wherein:
X is a matrix of two different clustering methods;
β is a coefficient representing a seasonality of two or more clusters, calculated using the following equation, where β 1 is the seasonality of a first cluster and β 2 is the seasonality of a second cluster:
β=(β 1 β 2 ) T .

19 . The at least one non-transitory memory storage module of claim 15 wherein:
the one or more clusters comprise at least a category-based cluster and a semantic-based cluster.

20 . The at least one non-transitory memory storage module of claim 15 wherein:
generating the demand forecast for the SKU comprises using a moving average technique to estimate the demand forecast.

21 . The at least one non-transitory memory storage module of claim 15 wherein:
calculating the cluster seasonality profile for each of the one or more clusters comprises:

using a mixed-effect model to remove randomness of cluster data to calculate the cluster seasonality profile.

Assignees

• Wal Mart Stores Inc

Inventors

• Feng Huijun
• Ray Shubhankar
• Jha Abhay

Classifications

• G06Q10/087 Primary

  Inventory or stock management, e.g. order filling, procurement or balancing against orders · CPC title

• G06Q30/0202

  Market predictions or forecasting for commercial activities · CPC title

• G06Q10/06315 Primary

  Needs-based resource requirements planning or analysis · CPC title

• G06Q30/0201

  Market modelling; Market analysis; Collecting market data · CPC title

• Y04S10/50

  Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications · CPC title