Causal Relation Model Pseudo-Cause Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current causal relation learning methods, such as statistical independence-based and score-based methods, suffer from low accuracy in discovering causal relations among variables, leading to incorrect decision-making in fields like product retail, healthcare, and software development due to retained false causal relations.
Innovation Solution
A data processing method that optimizes causal relation models by using score-based learning and independence checks to remove pseudo-causes, synthesizing both methods to enhance accuracy and provide a more accurate understanding of complex mechanisms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If statistical independence-based method or score-based method is used to discover causal relations, then the causal relation model can be obtained, but the accuracy is low due to retained false causal relations
Solution Approach 1:
The patent combines statistical independence-based method and score-based method into a unified causal discovery framework. The statistical independence test is used to identify potential causal relations, while the score-based evaluation ranks and selects the most reliable causal relations, thereby improving both accuracy and reliability by leveraging the strengths of both approaches
Solution Approach 2:
The patent implements a feedback mechanism where the causal relation model is continuously refined by comparing predicted causal relations with actual data patterns. The independence test results feed into the score-based selection, and the resulting model is validated against the original data, creating an iterative improvement process that enhances accuracy while maintaining reliability
Data Source
AI summary
Embodiments of the present disclosure provide a data processing method, an electronic device and a computer-readable storage medium. The data processing method comprises: obtaining a model representing causal relations among a plurality of variables based on a set of observation data of the plurality of variables; determining, based on the obtained model, a first and a second variables having direct causal relation in the plurality of variables; determining whether the first and second variables are independent from each other; and in response to the first and second variables being independent from each other, deleting the direct causal relation between the first and second variables from the obtained model. With the data processing method of the present disclosure, pseudo-causes can be removed effectively so that causal relations among a plurality of variables can be represented more accurately.