US 12,468,719 B2
Time series forecasting
Xi Cheng, Kirkland, WA (US); Amir H. Hormati, Seattle, WA (US); Lisa Yin, Redmond, WA (US); and Umar Syed, Edison, NJ (US)
Assigned to Google LLC, Mountain View, CA (US)
Filed by Google LLC, Mountain View, CA (US)
Filed on May 25, 2023, as Appl. No. 18/323,766.
Application 18/323,766 is a continuation of application No. 16/986,861, filed on Aug. 6, 2020, granted, now 11,693,867.
Claims priority of provisional application 63/026,573, filed on May 18, 2020.
Prior Publication US 2023/0297583 A1, Sep. 21, 2023
Int. Cl. G06F 16/2458 (2019.01); G06F 16/22 (2019.01)
CPC G06F 16/2477 (2019.01) [G06F 16/221 (2019.01); G06F 16/2282 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method executed by data processing hardware that causes the data processing hardware to perform operations comprising:
receiving a time series forecasting query requesting a time series forecast, the time series forecast comprising a forecast of future data based on current data;
training a plurality of machine learning models for the time series forecast using an order of a moving-average model, wherein:
each respective machine learning model of the plurality of machine learning models is trained on the same data blocks of the current data using a different order of the moving-average model; and
each machine learning model of the plurality of machine learning models is trained to predict the time series forecast based on the same data blocks of the current data;
for each respective model of the plurality of machine learning models, determining, using the current data, a respective prediction error of the respective machine learning model;
selecting, using the respective prediction error of each respective machine learning model, a single one of the plurality of machine learning models that best fits the respective time series forecast;
forecasting the future data based on the selected best fitting machine learning model and the current data; and
returning the forecasted future data for the time series forecast requested by the time series forecasting query.