Time Series Prediction Model Segmentation for Peak Value Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing prediction methods face significant challenges in accurately forecasting peak values of objective variables, particularly in long-term predictions, due to large prediction errors at extreme values.
Innovation Solution
An information processing device that divides time series data into sections based on objective variable values, generates prediction models for each section using explanatory and objective variables, and selects the appropriate model for accurate peak value prediction, incorporating techniques like Kernel density estimation and model learning methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single prediction model is used for the entire time series data, then the device complexity is low, but the prediction accuracy at peak values deteriorates
Solution Approach 1:
The patent divides the time series data into multiple sections based on the values of the objective variable, creating distinct segments for different value ranges. This segmentation allows the system to apply appropriate prediction models to each section, improving prediction accuracy at peak values while managing complexity through structured division of the data space.
Solution Approach 2:
The patent generates different prediction models tailored to specific sections of the time series data. Each section receives a customized model trained on its specific characteristics, particularly optimizing for peak value prediction in relevant sections. This local customization improves prediction accuracy without requiring a completely complex system, as each local model focuses on its specific data range.
2Duration of action of moving object
If the prediction period is extended, then the forecast coverage is improved, but the prediction error at peak values increases
Solution Approach 1:
The patent segments the prediction task into multiple sections based on objective variable values. For long-term predictions, the system can identify which section the predicted values will fall into and apply the appropriate specialized model, maintaining accuracy even over extended prediction periods where traditional single-model approaches would fail.
Solution Approach 2:
The patent performs preliminary classification of data points into different sections based on their value ranges before prediction. This preliminary action allows the system to pre-select the appropriate model for each potential prediction outcome, ensuring that even long-term forecasts land in the correct model section and maintain accuracy.
Data Source
AI summary
According to one embodiment, an information processing device includes: a divider configured to divide time series data of an objective variable into a plurality of first sections based on values of the objective variable; a model generator configured to generate, based on time series data of an explanatory variable and the time series data of the objective variable, a plurality of prediction models in which the explanatory variable and the objective variable are associated, for the plurality of first sections; a selector configured to select a first section from the plurality of first sections based on at least one of the time series data of the explanatory variable and the time series data of the objective variable; and a predictor configured to predict the value of the objective variable by using the prediction model generated for the selected first section.