Supply chain demand uncensoring

Patent metadata

Field	Value
Publication number	US-12387170-B2
Application number	US-202117566739-A
Country	US
Kind code	B2
Filing date	Dec 31, 2021
Priority date	Dec 31, 2021
Publication date	Aug 12, 2025
Grant date	Aug 12, 2025

Abstract

A processor may estimate uncensored demand from historical supply chain data. The processor may ingest historical data. The processor may convert the historical data to a dataset of multiple time series corresponding to sales for different products and locations and channels across multiple time points that is usable by an uncensored demand estimation machine learning model. The processor may train the uncensored demand estimation machine learning model by applying optimization solver techniques for deep learning.

First claim

What is claimed is:

1. A computer implemented method to estimate uncensored demand from historical supply chain data, the method comprising:
ingesting, by a processor, historical data;
converting the historical data to a dataset of multiple time series corresponding to sales for different products and locations and channels across multiple time points, wherein the dataset of multiple time series is usable by an uncensored demand estimation machine learning model;
training the uncensored demand estimation machine learning model with the dataset of multiple time series; and
estimating, with the uncensored demand estimation machine learning model, an uncensored demand.

2. The computer implemented method of claim 1 , further comprising:
incorporating, automatically, cross-time-series information from the historical data as part of training the model;
applying the trained uncensored demand estimation machine learning model to sales data for an object, wherein the sales data includes data associated with demand censoring; and
outputting an uncensored demand for the object.

3. The computer implemented method of claim 2 , wherein the sales data is associated with a time unit, and wherein the uncensored demand is associated with a future time unit.

4. The computer implemented method of claim 2 , wherein the uncensored demand for the object includes an uncensored demand probability distribution.

5. The computer implemented method of claim 2 , wherein the sales data includes temporal time-series data associated with sales of the object.

6. The computer implemented method of claim 2 , further comprising a method to estimate sales realization, the method comprising:
determining additional demand for a time period by taking the difference between an aggregate uncensored demand for the time period and an aggregate observed demand for the time period; and
allocating the additional demand to a time unit of the time period based on the uncensored demand probability distribution associated with the time unit.

7. The computer implemented method of claim 1 , the method further comprising:
training the uncensored demand estimation machine learning model using a factor dropout dataset to enable predicting uncensored demand for time series with new attribute values.

8. A system comprising:
a memory; and
a processor in communication with the memory, the processor being configured to perform operations comprising:
ingesting historical data;
converting the historical data to a dataset of multiple time series corresponding to sales for different products and locations and channels across multiple time points, wherein the dataset of multiple time series is usable by an uncensored demand estimation machine learning model;
training the uncensored demand estimation machine learning model with the dataset of multiple time series; and
estimating, with the uncensored demand estimation machine learning model, an uncensored demand.

9. The system of claim 8 , the processor being further configured to perform operations comprising:
incorporating, automatically, cross-time-series information from the historical data as part of training the model;
applying the trained uncensored demand estimation machine learning model to sales data for an object, wherein the sales data includes data associated with demand censoring; and
outputting an uncensored demand for the object.

10. The system of claim 9 , wherein the sales data is associated with a time unit, and wherein the uncensored demand is associated with a future time unit.

11. The system of claim 9 , wherein the uncensored demand for the object includes an uncensored demand probability distribution.

12. The system of claim 9 , wherein the sales data includes temporal time-series data associated with sales of the object.

13. The system of claim 9 , the processor being further configured to perform operations comprising:
determining additional demand for a time period by taking the difference between an aggregate uncensored demand for the time period and an aggregate observed demand for the time period; and
allocating the additional demand to a time unit of the time period based on the uncensored demand probability distribution associated with the time unit.

14. The system of claim 8 , the processor being further configured to perform operations comprising:
training the uncensored demand estimation machine learning model using a factor dropout dataset to enable predicting uncensored demand for time series with new attribute values.

15. A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to perform operations, the operations comprising:
ingesting historical data;
converting the historical data to a dataset of multiple time series corresponding to sales for different products and locations and channels across multiple time points, wherein the dataset of multiple time series is usable by an uncensored demand estimation machine learning model;
training the uncensored demand estimation machine learning model with the dataset of multiple time series; and
estimating, with the uncensored demand estimation machine learning model, an uncensored demand.

16. The computer program product of claim 15 , the processor being further configured to perform operations comprising:
incorporating, automatically, cross-time-series information from the historical data as part of training the model;
applying the trained uncensored demand estimation machine learning model to sales data for an object, wherein the sales data includes data associated with demand censoring; and
outputting an uncensored demand for the object.

17. The computer program product of claim 16 , wherein the sales data is associated with a time unit, and wherein the uncensored demand is associated with a future time unit.

18. The computer program product of claim 16 , wherein the uncensored demand for the object includes an uncensored demand probability distribution.

19. The computer program product of claim 16 , wherein the sales data includes temporal time-series data associated with sales of the object.

20. The computer program product of claim 16 , the processor being further configured to perform operations comprising:
determining additional demand for a time period by taking the difference between an aggregate uncensored demand for the time period and an aggregate observed demand for the time period; and
allocating the additional demand to a time unit of the time period based on the uncensored demand probability distribution associated with the time unit.

Assignees

• IBM

Inventors

• Quanz Brian Leo
• Harsha Pavithra
• Shah Dhruv
• Ramakrishna Mahesh
• Koc Ali

Classifications

• G06Q10/06315

  Needs-based resource requirements planning or analysis · CPC title

• G06Q10/06375 Primary

  Prediction of business process outcome or impact based on a proposed change · CPC title

• G06Q10/04

  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) · CPC title

• G06Q10/087 Primary

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