Interpretable User Segmentation Rules via Predictive Model Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user segmentation approaches, particularly a posteriori methods, achieve high accuracy but suffer from interpretability issues, making it difficult for operators to understand and configure computing systems effectively, and they can be computationally inefficient when dealing with large interaction data sets.
Innovation Solution
A predictive model that generates interpretable rules by analyzing user interaction data points, selecting an optimal set of rules based on both accuracy and interpretability parameters, and presenting these rules on a user interface to facilitate understanding and efficient computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a posteriori segmentation using predictive models (e.g., deep learning systems) is used, then accuracy of user segmentation is improved, but interpretability deteriorates
Solution Approach 1:
The patent segments the predictive model into two distinct components: an black-box predictive model for accurate segmentation and a separate rule generation module that creates interpretable rules. This segmentation allows each component to specialize - the predictive model achieves high accuracy while the rule generation module provides interpretability, resolving the contradiction between accuracy and interpretability.
Solution Approach 2:
The patent introduces an intermediary rule generation module that translates the outputs of the predictive model into human-readable rules. This intermediary component acts as a bridge between the accurate but opaque predictive model and the need for interpretable results, allowing operators to understand the segmentation logic without sacrificing accuracy.
2Manufacturing precision
If complex predictive models are used to improve segmentation accuracy, then manufacturing precision is improved, but device complexity increases
Solution Approach 1:
The system segments the overall solution into a complex predictive model component and a simpler rule generation component. This allows the complex model to focus solely on accurate prediction while the rule generation component handles interpretability, distributing the complexity burden and making the overall system more manageable despite using sophisticated models.
Solution Approach 2:
The patent creates a simplified copy or representation of the predictive model's logic in the form of human-readable rules. Instead of directly presenting the complex model structure, it generates an approximate copy in rule form that is easier to understand and work with, reducing the perceived complexity for operators.
3Measurement precision
If large amounts of interaction data are analyzed to improve segmentation quality, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent performs preliminary actions by pre-processing the interaction data and pre-training the predictive model offline before deployment. This allows the model to learn from large datasets in advance, so that during actual operation, segmentation can be performed quickly using the already-trained model, reducing the time loss during critical operations.
Solution Approach 2:
The patent extracts the essential segmentation patterns from large datasets during the offline training phase, separating the heavy computational work from the operational phase. This extraction of key insights during preliminary processing allows for fast execution during actual use, reducing time loss when segmentation is needed.
Data Source
AI summary
Various embodiments describe user segmentation. In an example, potential rules are generated by applying a frequency-based analysis to user interaction data points. Each of the potential rules includes a set of attributes of the user interaction data points and indicates that these data points belong to a segment of interest. An objective function is used to select an optimal set of rules from the potential rules for the segment of interest. The potential rules are used as variable inputs to the objective function and this function is optimized based on interpretability and accuracy parameters. Each rule from the optimal set is associated with a group of the segment of interest. The user interaction data points are segments into the groups by matching attributes of these data points with the rules.