Data Analysis Device for Feature Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data mining techniques require trial and error to design optimal features, making the process time-consuming and uncertain about the direction and extent of improvement or deterioration in objective variables when feature values change.
Innovation Solution
A data analyzing device that generates a prediction model for objective variables based on feature values, allowing users to simulate changes in objective variables by adjusting feature values, thereby determining the propriety of measures before implementation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic feature generation is performed by applying arithmetic operators to original features, then a large amount of new features can be generated, but it remains unclear how the objective variable changes when feature values change
Solution Approach 1:
The system displays the relationship between feature values and objective variable values, providing feedback to users about how changes in feature values affect the objective variable. This allows users to understand the impact of automatic feature generation on the target outcome.
Solution Approach 2:
A display unit is introduced as an intermediary between the automatic feature generation process and the user. It visualizes the relationship between feature values and objective variable values, making the otherwise invisible impact of feature changes observable and understandable.
2Measurement precision
If trial and error is performed by experienced analysts to design optimal features, then analysis accuracy can be improved, but the analysis process takes a long time
Solution Approach 1:
The system performs preliminary feature generation and displays the relationship between feature values and objective variable values before final analysis. This allows users to pre-evaluate potential features and make informed decisions, reducing the need for time-consuming trial and error iterations.
Solution Approach 2:
The system automatically generates features and displays their relationships with objective variables, enabling users to independently evaluate and select optimal features without requiring extensive expert trial and error, thus reducing analysis time while maintaining accuracy.
3Quantity of substance
If comprehensive arithmetic operators are applied to generate new features, then feature quantity increases, but the complexity of the analysis system increases
Solution Approach 1:
The display unit serves as an intermediary that simplifies the complex relationship between numerous generated features and the objective variable by visualizing it in an understandable format. This allows the system to handle large numbers of features without proportionally increasing user-facing complexity.
Data Source
AI summary
To make it possible to determine in advance the propriety of a measure by simulating a change of an objective variable due to a change of a value of a feature value. A prediction model for predicting an objective variable is generated based on feature values of a plurality of attributes included in analysis target data. An adjustment of a representative value of a feature value is received based on a user's operation input. The feature values of the plurality of attributes are changed according to adjustment amount of the representative value, and the numeric value of the objective variable is recalculated from the prediction model and displayed on a monitor.