Time Series Prediction Model Using Derivative Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional technologies face challenges in accurately modeling physical phenomena due to errors in long-term future prediction and the strong influence of certain data points, particularly in temperature prediction and computational fluid dynamics, where data weighting and normalization are difficult to manage effectively.
Innovation Solution
An information processing device that calculates time derivative values and fluctuation differences of variables, using machine learning to estimate coefficients for a linear regression equation, incorporating variance-covariance matrices to address error accumulation and data weighting, and mixing short-term and long-term components for improved prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional machine learning is used to model physical phenomena from time-series data, then the model generation process is simple, but the accuracy of long-term future prediction deteriorates due to error accumulation
Solution Approach 1:
The patent segments the time-series data into multiple intervals and performs machine learning separately for each interval. By dividing the long-term prediction task into shorter segments, error accumulation is prevented while maintaining prediction accuracy across the entire time range. Each segment's model is trained independently on its specific time interval data.
2Reliability
If data normalization is applied to manage data weighting, then the influence of certain data points is reduced, but the complexity of data processing increases
Solution Approach 1:
The patent applies data normalization as a preliminary action before performing machine learning. By normalizing the time-series data in advance, the influence of data points with different scales is balanced, preventing any single data point from dominating the model training. This preliminary processing simplifies subsequent analysis while ensuring reliable data weighting.
3Measurement precision
If variance-covariance matrices are incorporated to address error accumulation, then prediction accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent dynamically adjusts the application of variance-covariance matrices based on the specific time interval and data characteristics. Rather than applying a fixed complex computational framework throughout, the method adapts the use of variance-covariance analysis to where it is most needed, balancing computational complexity with prediction accuracy improvements.
Data Source
AI summary
According to one embodiment, an information processing device includes a memory and one or more processors. The memory stores time-series data including one or more variables. The one or more processors are coupled to the memory and configured to: calculate a time derivative value of each of the variables; calculate a difference indicating fluctuation of a long-term component of the corresponding variable based on a designated time sample interval; estimate a coefficient of a linear regression equation by machine learning using the time derivative value and the difference as learning data; and output the linear regression equation.