Title:
データ予測システム及びその方法
Document Type and Number:
Japanese Patent JP7257276
Kind Code:
B2
Abstract:
To provide a data prediction system capable of reducing errors for prediction values.SOLUTION: A data prediction system includes a factor fluctuation evaluation section 251 and a prediction section 252, and calculates prediction data. The factor fluctuation evaluation section 251 outputs data for controlling a contribution level for each factor when identifying a prediction model. The contribution level for each factor is shifted based on a fluctuation amount of each factor. The prediction section 252 identifies the prediction model based on a specimen of a prediction subject, a specimen of the factor, and the data for controlling the contribution level for each factor. A prediction value for the prediction target is outputted by the identified prediction model. Thus, the prediction section 252 identifies the prediction model that can be computationally processed based on the factor, and calculates the prediction value for the prediction subject associated with the factor. Accordingly, the prediction section 252 identifies the prediction model based on an observation specimen of the prediction subject, an observation specimen of the factor, and the data for controlling the contribution level for each factor, and outputs the prediction value for the prediction subject.SELECTED DRAWING: Figure 5
Inventors:
Masato Utsumi
Toru Watanabe
Ikuo Shigemori
Isao Hagiya
Toru Watanabe
Ikuo Shigemori
Isao Hagiya
Application Number:
JP2019126321A
Publication Date:
April 13, 2023
Filing Date:
July 05, 2019
Export Citation:
Assignee:
株式会社日立製作所
International Classes:
G06Q10/04; G06Q30/0202; H02J3/00
Domestic Patent References:
| JP2018092439A | ||||
| JP2018163515A | ||||
| JP2019049850A | ||||
| JP2017224268A | ||||
| JP2018124727A | ||||
| JP2016127643A |
Attorney, Agent or Firm:
Patent Attorney Corporation Sunnext International Patent Office