Time series data prediction apparatus and time series data prediction method

Patent metadata

Field	Value
Publication number	US-2021264375-A1
Application number	US-202017009204-A
Country	US
Kind code	A1
Filing date	Sep 1, 2020
Priority date	Feb 25, 2020
Publication date	Aug 26, 2021
Grant date	—

Abstract

Prediction of time series data can be performed while coping with a “non-equidistant event,” “extrapolation,” and a “base line shift.” An event regression section calculates event prediction data that is predicted values of metric data in an intended period including a past certain period on the basis of actual measured value data indicating values of the metric data in the past metric data. A correction section calculates, as prediction result data that is a prediction result of the time series data, data obtained by shifting each value of the event prediction data in response to a difference between the actual measured value data and the event prediction data in a same period.

First claim

1 . A time series data prediction apparatus comprising:
a memory for storing data; an input/output device; and a processor, operatively coupled to and in communication with the memory and the input/output device, the processor: receives information from the input/output device, calculates event prediction data that is predicted values of time series data in an intended period including a past certain period on a basis of actual measured value data indicating values of the time series data in the past certain period, the data including information relating to events occurring in unequal time intervals; and calculates, as prediction result data that is a prediction result of the time series data, data obtained by shifting each value of the event prediction data in response to a difference between the actual measured value data and the event prediction data in a same period.

2 . The time series data prediction apparatus according to claim 1 , wherein the processor:
calculates trend prediction data indicating a tendency of the time series data in the intended period on the basis of the actual measured value data; and calculates relative value data indicating relative values of the actual measured value data to the trend prediction data in the certain period, wherein the processor: calculates the event prediction data on a basis of the relative value data, and shifts each value of the event prediction data using the trend prediction data as the difference.

3 . The time series data prediction apparatus according to claim 2 , wherein the processor:
generates a trend prediction model with elapsed time assumed as explanatory variables and values of the time series data assumed as objective variables on the basis of the actual measured value data, and calculates the trend prediction data using the trend prediction model.

4 . The time series data prediction apparatus according to claim 3 , wherein
the trend prediction model is a linear regression model.

5 . The time series data prediction apparatus according to claim 2 , wherein the processor:
generates an event prediction model with calendar information assumed as explanatory variables and values of the time series data assumed as objective variables on the basis of the relative value data, and calculates the event prediction data using the event prediction model.

6 . The time series data prediction apparatus according to claim 5 , wherein
the event prediction model includes a decision tree model.

7 . The time series data prediction apparatus according to claim 6 , wherein
the decision tree model includes a plurality of decision trees each calculating candidate data that serves as a candidate of the event prediction data, and the processor calculates, as values of the event prediction data, a lower limit, a representative value, and an upper limit on a basis of a plurality of pieces of candidate data calculated by the decision trees.

8 . The time series data prediction apparatus according to claim 1 , wherein the processor:
shifts each value of the event prediction data in and after the certain period in response to the difference between the actual measured value data and the event prediction data in a terminal period including an end of the certain period.

9 . The time series data prediction apparatus according to claim 2 , wherein the processor:
repeatedly calculates the event prediction data while shifting the certain period, and determines whether the difference between the actual measured value data and the event prediction data in the terminal period including the end of the certain period is equal to or greater than a certain value whenever the event prediction data is calculated, shifts each value of the event prediction data in response to the difference in a case in which the difference is equal to or greater than the certain value and the event prediction data is finally calculated event prediction data, and changes an approach of calculating the trend prediction data by the trend calculation section in a case in which the difference is equal to or greater than the certain value and the event prediction data is not the finally calculated event prediction data.

10 . A time series data prediction method by a time series data prediction apparatus, the time series data prediction method comprising:
calculating, by a processor operatively coupled to and in communication with a memory and an input/output device, based upon information received from the input/output device, event prediction data that is predicted values of time series data in an intended period including a past certain period on a basis of actual measured value data indicating values of the time series data in the past certain period, the data including information relating to events occurring in unequal time intervals; and calculating, by the processor, as prediction result data that is a prediction result of the time series data, data obtained by shifting each value of the event prediction data in response to a difference between the actual measured value data and the event prediction data in a same period.

Assignees

• Hitachi Ltd

Inventors

• Himura Yosuke
• Masuda Mineyoshi

Classifications

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06Q10/04 Primary

  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

• G06Q10/0631

  Resource planning, allocation, distributing or scheduling for enterprises or organisations · CPC title

• G06N20/00

  Machine learning · CPC title

• G06Q10/1093 Primary

  Calendar-based scheduling for persons or groups · CPC title