# US 2023/0274162 A1 · Demand forecasting system, learning system, and demand forecasting method

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2023/0274162 A1
- Document ID: US20230274162A1
- Status: Published application
- Kind: A1
- Application number: 18/071,720
- Filed: 2022-11-30
- Published: 2023-08-31
- Earliest claimed: 2022-02-25
- Assignees: Toyota Jidosha Kabushiki Kaisha
- Applicants: Toyota Jidosha Kabushiki Kaisha
- Inventors: Tomotaka SUGAI; Yoshiumi KAWAMURA; Makoto TANIGUCHI
- Web page: https://patentgrep.com/patent/US20230274162A1

## CPC

- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06Q 10/04: 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)
- Y02P 90/30: Computing systems specially adapted for manufacturing
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- Y02P: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS

## Abstract

A demand forecasting system obtains first forecasting data indicating a representative demand forecasting value of a provision target for each cluster by executing a time series clustering process, extracting trend component data for each cluster, and calculating pattern data. The demand forecasting system obtains second forecasting data that is data indicating a forecasting value of a difference for each provision target and obtains third forecasting data indicating a demand forecasting value for each provision target.

## Claims

### Claim 1

A demand forecasting system comprising a processor configured to:
input, regarding a provision target that is a commodity or a service, actual result data including the number of time series actual demand results for each of a plurality of types of provision targets and execute a time series clustering process on the input actual result data for each provision target;
execute, for each cluster classified by the time series clustering process, an extraction process of extracting trend component data of the cluster from the actual result data;
execute, for each cluster, a pattern calculation process of calculating pattern data indicating a demand pattern for a predetermined period by performing a statistical process on first data that is data remaining after the trend component data is extracted from the actual result data for the predetermined period;
execute, for each cluster, a first forecasting process of obtaining first forecasting data indicating a representative demand forecasting value of the provision target belonging to the cluster for each cluster by calculating a sum of the pattern data and the trend component data;
input, for each provision target, a difference between the actual result data of the provision target and the first forecasting data for the cluster to which the provision target belongs, in a learning model in which a regression learning is performed on the difference for the provision target, and execute a second forecasting process of obtaining second forecasting data that is data indicating a forecasting value of the difference for each provision target; and
obtain third forecasting data that is data indicating a demand forecasting value for each provision target by calculating a sum of the first forecasting data and the second forecasting data for each provision target.

### Claim 2 (depends on 1)

The demand forecasting system according to claim 1, wherein:
the processor is configured to
execute, for each cluster, a determination process of determining whether a trend component is included in extracted data that is data obtained after the trend component data is extracted in the extraction process, and
execute, when a determination is made that a trend component is included for at least one of the clusters as a result of the determination process, the time series clustering process and the extraction process on the extracted data instead of the actual result data, until a determination is made in the determination process that a trend component is not included; and
the pattern calculation process uses data remaining after the trend component data is extracted from the extracted data instead of the actual result data as the first data at a stage when a determination is made in the determination process that a trend component is not included, and calculates the pattern data by performing the statistical process on the first data for the predetermined period.

### Claim 3 (depends on 1)

The demand forecasting system according to claim 1, wherein:
the predetermined period is a period in which start date and end date are represented in days or months within a year; and
the statistical process is a process of calculating a statistical value of the first data for the predetermined period in the first data for a predetermined number of years, for each cluster.

### Claim 4 (depends on 1)

The demand forecasting system according to claim 1, wherein the first forecasting process includes an approximation process of approximating the trend component data, which is extracted in the extraction process, to a polynomial, and is a process of obtaining the first forecasting data for each cluster by calculating a sum of the pattern data and data indicated by the polynomial for each cluster.

### Claim 5 (depends on 1)

The demand forecasting system according to claim 1, wherein data after standardization for the types of provision targets is used in the time series clustering process, the extraction process, and the pattern calculation process as the actual result data, and data before the standardization is used in the second forecasting process as the actual result data.

### Claim 6 (depends on 1)

The demand forecasting system according to claim 1, wherein the provision targets are any one of
parts that constitute one or a plurality of products,
predetermined quantities of materials that are used in a case of manufacturing one or a plurality of products,
products that belong to one or a plurality of product groups, and
services that belong to one or a plurality of service groups.

### Claim 7

A learning system comprising a processor configured to:
input, regarding a provision target that is a commodity or a service, a difference between actual result data including the number of time series actual demand results for each of a plurality of types of provision targets and first forecasting data that is data indicating a representative demand forecasting value, which is obtained for each cluster classified by executing a time series clustering process on the actual result data for each provision target, for a cluster to which the provision target belongs, in a regression learning model; and
execute a regression learning of the regression learning model and update the regression learning model such that the regression learning model becomes a model that inputs the difference and outputs a forecasting value of the difference for a predetermined period during operation,
wherein the first forecasting data is obtained, as data indicating a representative demand forecasting value of the provision target belonging to the cluster for each cluster,
by executing, for each cluster, an extraction process of extracting trend component data of the cluster from the actual result data,
by executing, for each cluster, a pattern calculation process of calculating pattern data indicating a demand pattern for a predetermined period by performing a statistical process on first data that is data remaining after the trend component data is extracted from the actual result data for the predetermined period, and
by executing, for each cluster, a first forecasting process of calculating a sum of the pattern data and the trend component data.

### Claim 8 (depends on 7)

The learning system according to claim 7, wherein:
the predetermined period is a period in which start date and end date are represented in days or months within a year; and
the statistical process is a process of calculating a statistical value of the first data for the predetermined period in the first data for a predetermined number of years, for each cluster.

### Claim 9 (depends on 7)

The learning system according to claim 7, wherein data after standardization for the types of provision targets is used in the time series clustering process, the extraction process, and the pattern calculation process as the actual result data, and a difference, which is calculated by using data before the standardization, is used for the difference, which is input to the regression learning model, as the actual result data.

### Claim 10 (depends on 7)

The learning system according to claim 7, wherein the provision targets are any one of
parts that constitute one or a plurality of products,
predetermined quantities of materials that are used in a case of manufacturing one or a plurality of products,
products that belong to one or a plurality of product groups, and
services that belong to one or a plurality of service groups.

### Claim 11

A demand forecasting method comprising:
inputting, regarding a provision target that is a commodity or a service, actual result data including the number of time series actual demand results for each of a plurality of types of provision targets and executing a time series clustering process on the input actual result data for each provision target;
executing, for each cluster classified by the time series clustering process, an extraction process of extracting trend component data of the cluster from the actual result data;
executing, for each cluster, a pattern calculation process of calculating pattern data indicating a demand pattern for a predetermined period by performing a statistical process on first data that is data remaining after the trend component data is extracted from the actual result data for the predetermined period;
executing, for each cluster, a first forecasting process of obtaining first forecasting data indicating a representative demand forecasting value of the provision target belonging to the cluster for each cluster by calculating a sum of the pattern data and the trend component data;
inputting, for each provision target, a difference between the actual result data of the provision target and the first forecasting data for the cluster to which the provision target belongs, in a learning model in which a regression learning is performed on the difference for the provision target, and executing a second forecasting process of obtaining second forecasting data that is data indicating a forecasting value of the difference for each provision target; and
obtaining third forecasting data that is data indicating a demand forecasting value for each provision target by calculating a sum of the first forecasting data and the second forecasting data for each provision target.
