Parameter Selection for Time Series Prediction Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for adjusting parameters in time series data prediction models, such as those used in manufacturing sites handling hazardous substances, are labor-intensive and prone to determining improper parameter sets due to the inclusion of abnormal data.
Innovation Solution
An information processing apparatus that calculates index values for adjustment parameter sets and selects an appropriate set based on these values, as well as those of neighboring sets, to avoid the impact of abnormal data, using a concept of 'distance' in time position to evaluate parameter sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If parameter adjustment is performed through repetition of trials and errors based on prediction trends and engineer knowledge, then parameter adjustment can be carried out with available data, but the work takes a lot of labor and time
Solution Approach 1:
The patent replaces the manual trial-and-error parameter adjustment process with an AI-based automated system. The AI model automatically evaluates multiple parameter sets and selects optimal ones, substituting the mechanical manual process with an intelligent automated system that reduces labor and time requirements while maintaining or improving adjustment accuracy.
2Loss of time
If existing AI methods are used for parameter adjustment, then labor and time requirements are reduced, but improper parameter sets may be determined due to inclusion of abnormal data such as outliers
Solution Approach 1:
The patent extracts and removes abnormal data points (outliers) from the dataset before parameter adjustment. By identifying and excluding these abnormal data points, the system prevents them from influencing the parameter selection process, thereby ensuring that only reliable and accurate parameter sets are determined while maintaining the efficiency benefits of AI-based automation.
3Device complexity
If parameter adjustment is performed without considering the temporal position of data, then evaluation can be simplified, but abnormal data at different time positions can improperly influence parameter determination
Solution Approach 1:
The patent segments the time series data into distinct normal and abnormal sections based on temporal position. By dividing the data into these segments and evaluating parameter sets separately for each segment, the system can identify parameter sets that perform well across different time periods without being unduly influenced by abnormal data points, thereby improving selection accuracy while maintaining manageable complexity.
Data Source
AI summary
An information processing apparatus includes an index value calculation module and a parameter determination module. The index value calculation module calculates values of preset indexes for each of adjustment parameter sets to calculate regression coefficients which are used to derive a predicted value from past data. The parameter determination module selects an appropriate adjustment parameter set based on the index values calculated for each of the adjustment parameter sets and index values of neighborhood adjustment parameter sets with respect to the interested each adjustment parameter set.