/Patent Tracker
GooglePatent applicationAI & softwareInterest 4/10

Google patents method to speed up complex data forecasting

Patent drawing
Patent drawing from the official filing.

This filing describes a way to predict future trends in data that involves multiple connected factors, such as weather and sales. It breaks the problem into smaller, independent parts that can be calculated at the same time to make the process faster.

Why it mattersIt offers a more efficient way to handle complex predictions, which could improve the speed of internal data analysis tools.

In the filing’s own words

Systems and methods described herein can provide for parallelized multivariate time series forecasting. The systems and methods described herein can provide for obtaining an input dataset including a plurality of univariate time series for a plurality of endogenous variables. Fitted regression models may be identified for each univariate time series by parallelized independent regression. A plurality of residual series may be determined, and one or more exogenous feature values may be extracted based on the fitted regression models. The systems and methods described herein can provide for forecasting a plurality of residual series using fitted forecasting models. Model parameters, forecasted residual series, and exogenous feature values may be stored in a multivariate model table for each endogenous variable. Forecasted time series data can then be generated using this table. The parallelized independent regression on each univariate time series may comprise determining regression matrices respective to a time series identifier. The systems and methods described herein can provide for generating forecasted time series data for each of the endogenous variables.

What exactly they claim (claim 1)

1. A computing system for parallelized multivariate time series forecasting, comprising: one or more processors; one or more non-transitory storage media coupled to the one or more processors, the one or more non-transitory storage media storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, the operations including: obtaining an input dataset comprising a plurality of univariate time series respective to a plurality of endogenous variables; identifying a plurality of fitted regression models respective to the plurality of univariate time series by parallelized independent regression on each univariate time series of the plurality of univariate time series; determining a plurality of residual series respective to the plurality of univariate time series and extracting one or more exogenous feature values from the plurality of univariate time series based on the plurality of fitted regression models; forecasting a plurality of forecasted residual series respective to each of the plurality of residual series by a respective fitted residual forecasting model of a plurality of fitted forecasting model; storing in a multivariate model table, respective to each endogenous variable: parameters of the respective fitted residual forecasting model; a respective forecasted residual series of the plurality of forecasted residual series; and the one or more exogenous feature values; and generating, using the multivariate model table, forecasted time series data for each of the endogenous variables.

Facts

Company
Google (Google LLC)
Official title
Systems and Methods for Parallelized Multivariate Time Series Forecasting
Publication number
US20260320153A1
Application number
19176986
Filed
Apr 11, 2025
Published
Oct 15, 2026
Inventors
3
Claims
20

Classified under

G06F 17/18 Physics & computing
Read the full filing (official PDF) ↗ Open in the Patent Tracker

More from Google in AI & software

Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.