Factor Analysis Apparatus for Abnormal Event Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing factor analysis methods face difficulties in associating abnormal events with their underlying factors, especially when singularities or a lack of abnormalities are present in the data, making it challenging to derive relationships and predict future abnormalities.
Innovation Solution
A factor analysis apparatus that includes a similarity calculator, a first influence calculator, and a second influence calculator to associate target events with assumed factors by calculating degrees of similarity and influence, using techniques like Granger causality and term feature extraction, even in the absence of singularities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional factor analysis methods are used to associate abnormalities with factors, then clear association can be achieved when singularities or outliers are present in the data, but the method fails to effectively associate factors when no singularities or few abnormalities exist in the data
Solution Approach 1:
The patent transforms the analysis approach by changing from direct abnormality-based association to similarity-based association. It introduces new parameters including degree of similarity between data items and factors, degree of influence of data items on target events, and degree of influence of factors on target events. This parameter transformation enables effective factor analysis even when traditional singularities are absent in the data
Solution Approach 2:
The patent introduces an intermediary approach by using degree of similarity as a mediator between data items and factors. Instead of directly associating abnormalities with factors, the system calculates similarity degrees between data items and factors, then uses degree of influence as another intermediary to establish the relationship. This multi-layer intermediary approach enables factor association in situations where direct association fails
2Reliability
If more data items are analyzed to improve factor association in normal conditions, then better predictions can be made, but calculation complexity and processing time increase
Solution Approach 1:
The patent segments the factor analysis process into three distinct calculation stages: (1) calculating degree of similarity between each data item and each factor, (2) calculating degree of influence of each data item on the target event, and (3) calculating degree of influence of each factor on the target event using the previous results. This segmentation allows for systematic processing and enables parallel computation, reducing overall complexity while improving reliability
Data Source
AI summary
An apparatus as an aspect of the present invention is a factor analysis apparatus that analyzes a relationship between a target event that is a target of factor analysis and an assumed factor of the target event, and includes a similarity calculator, a first influence calculator, and a second influence calculator. The similarity calculator calculates a degree of similarity between a data item included in provided time-series data and the assumed factor. The first influence calculator calculates a first degree of influence indicating a degree of influence of the data item on the target event on the basis of time-series data of the data item and time-series data of the target event. The second influence calculator calculates a second degree of influence indicating a degree of influence of the assumed factor on the target event on the basis of the degree of similarity and the first degree of influence.