Automated Explanatory Variable Generation for Business Data Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data analysis methods for business management rely heavily on human intuition and experience, limiting the range of effective measures that can be derived from large amounts of business activity data, as they require a manager or analyst to establish hypotheses and verify them through data collection.
Innovation Solution
An integrated data analysis system that automatically generates a large number of explanatory variables using variable generation condition information, calculates correlations with objective variables, and displays results, enabling the identification of factors influencing business performance without relying on human intuition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional hypothesis-based analysis methods are used, then analysis can be performed with existing data, but the range of measures to be obtained is limited due to dependence on human ability
Solution Approach 1:
The analysis system automatically generates hypotheses and selects variables without human intervention. The system serves itself by autonomously performing hypothesis generation, variable selection, and correlation analysis, thereby expanding the range of measurable outcomes beyond human capability while maintaining manageable complexity through automation.
Solution Approach 2:
The system combines multiple data types (transaction data, sensor data, survey data) and integrates them with automated hypothesis generation capabilities. This composite approach merges diverse data sources and analytical functions into a unified system that overcomes the limitations of conventional single-method analysis.
2Productivity
If automated variable generation is implemented, then a large number of explanatory variables can be generated to expand analysis scope, but system complexity increases
Solution Approach 1:
The system replaces manual hypothesis generation and variable selection (mechanical human processes) with automated computational algorithms. This substitution enables high-speed, large-scale variable generation while reducing the operational complexity burden on users, as the system handles complexity internally through standardized automated procedures.
3Reliability
If manual hypothesis establishment by managers or analysts is used, then analysis aligns with human intuition and experience, but the establishment of hypothesis depends on human ability limiting the range of measures
Solution Approach 1:
The system incorporates feedback mechanisms where analysis results and correlation data are used to refine and generate new hypotheses iteratively. This feedback loop enables the system to learn from data patterns and continuously expand its analytical capabilities while maintaining reliability through data-driven validation of hypotheses.
Data Source
AI summary
A generation technique and an analysis technique of a large number of explanatory variables to derive effective measures by using various data are provided. Specifically, a factor which lurks in a large amount of data and affects business performance is identified by automatically generating a large number of explanatory variables and performing correlation analysis between the explanatory variables and an objective variable. Three operators representing condition, target, and arithmetic which are variable generation conditions are defined in advance for data inputted into an analysis system and a large number of explanatory variables are automatically generated by these operators.