Data Analyzing Device Feature Segmentation Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data mining techniques require extensive trial and error to design optimal features, making it time-consuming for analysts to understand which feature values influence objective variables and how they do so, especially when automatically generating new features.
Innovation Solution
A data analyzing device that automatically generates new feature values by applying predetermined functions to single or multiple attributes, divides these features into segments based on influence on objective variables, and calculates evaluation values to display which features impact the objective variable and how, allowing for in-depth analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic feature value generation is performed by applying predetermined functions to attributes, then productivity of data analysis is improved, but it becomes difficult to understand which feature values influence objective variables and how
Solution Approach 1:
The patent segments feature values into multiple ranges (e.g., low, medium, high) and calculates evaluation values for each segment separately. This segmentation allows the system to maintain automatic feature generation while providing interpretable information about which segments have the most influence on objective variables, resolving the contradiction between productivity and interpretability.
Solution Approach 2:
The patent introduces a feedback mechanism where evaluation values are calculated based on the relationship between segmented feature values and objective variables. This feedback provides analysts with actionable information about feature influence, allowing them to understand and refine the automatic feature generation process while maintaining high productivity.
2Manufacturing precision
If trial and error methods are used to design optimal features, then manufacturing precision of feature design is improved, but time consumption increases significantly
Solution Approach 1:
The patent performs preliminary segmentation of feature values into multiple ranges before conducting the full analysis. By pre-defining these segments and calculating evaluation values in advance, the system reduces the need for iterative trial and error while maintaining high feature design quality, thus resolving the time consumption issue.
Solution Approach 2:
The system performs automatic feature value generation and evaluation calculation without requiring manual trial and error by analysts. The automated process independently generates optimal features with high precision while significantly reducing time consumption, embodying the self-service principle.
3Adaptability or versatility
If comprehensive arithmetic operators are applied to generate new features, then adaptability of feature generation is improved, but device complexity increases
Solution Approach 1:
The patent implements a universal evaluation mechanism that works across different feature types and arithmetic operations. The same segmentation and evaluation value calculation process applies to all generated features regardless of how they were created, reducing system complexity while maintaining high adaptability for various feature generation scenarios.
Data Source
AI summary
To make it easier to perform an in-depth analysis by presenting to a user which feature value influences and how it influences an objective variable in a case of automatically generating feature values. For each of a plurality of feature values, by determining a division point indicating a change in influence on an objective variable, each feature value is divided into a plurality of segments and an evaluation value is calculated using an influence degree of each segment on the objective variable as an index. The segments of the feature values for which the evaluation values have been calculated are displayed on a monitor.