US 11,983,726 B2
Consumption prediction system and consumption prediction method
Wen-Kuang Chen, Taoyuan (TW); Chien-Kuo Hung, Taoyuan (TW); Chun-Hung Chen, Taoyuan (TW); and Chen-Chung Lee, Taoyuan (TW)
Assigned to QUANTA COMPUTER INC., Taoyuan (TW)
Filed by Quanta Computer Inc., Taoyuan (TW)
Filed on Aug. 24, 2020, as Appl. No. 17/000,681.
Claims priority of application No. 109110919 (TW), filed on Mar. 31, 2020.
Prior Publication US 2021/0303996 A1, Sep. 30, 2021
Int. Cl. G06Q 30/02 (2023.01); G06Q 30/0202 (2023.01); G06V 10/44 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06Q 50/12 (2012.01)
CPC G06Q 30/0202 (2013.01) [G06V 10/454 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06Q 50/12 (2013.01)] 9 Claims
OG exemplary drawing
 
1. A consumption prediction system, comprising:
a data storage device, configured to store historical environment data and a historical consumption record; and
a processor, configured to:
calculate a personal preference correlation coefficient;
input the historical environment data, the historical consumption record and the personal preference correlation coefficient into a first neural network model; wherein the first neural network model is used to generate a training model;
compare an accuracy rate of the training model to a training threshold;
responsive to a determination that the accuracy rate of the training model is higher than the training threshold, regard the training model as a prediction model;
input prediction environment data into the prediction model;
output, using the prediction model, a prediction result based on the prediction environment data;
extract a predicted guest number and a predicted turnover from the prediction result;
compare the predicted guest number with an actual guest number to generate a guest correction weight;
compare the predicted turnover with an actual turnover to generate a turnover correction weight;
determine whether the guest correction weight is higher than a guest weight threshold or whether the turnover correction weight is higher than a turnover weight threshold;
input the guest correction weight and the turnover correction weight into the first neural network model to adjust the training model according to the guest correction weight and the turnover correction weight in order to adjust the training model according to the guest correction weight and the turnover correction weight and update the training model, when the guest correction weight is higher than the guest weight threshold or the turnover correction weight is higher than the turnover weight threshold;
receive explicit data and store the explicit data in a first queue;
receive digital text data and store the digital text data in a second queue; and
receive picture data and store the picture data in a third queue.