Churn Factor Mining System for Network Applications
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for mining churned users in network applications are limited by their inability to handle stream data and require static samples, leading to a focus on macro-level effects and loss of original feature meanings during dimensionality reduction, which hinders specific optimization and improvement in user churn analysis.
Innovation Solution
A method and system that calculate churn factor influence weights based on user operation scenarios, identifying major churn factors by analyzing stream data and determining influence weights of features affecting user churn, allowing for detailed problem mining in network application submodules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dimensionality reduction processing is performed on features before modeling, then the multiple collinearity problem is avoided and model estimation accuracy is improved, but the original meanings of feature vectors are lost and specific function optimization becomes difficult
Solution Approach 1:
The patent segments the feature space by identifying and separating independent features from dependent features. It divides the analysis into two parts: first identifying independent features through statistical tests, then using only these independent features for modeling. This segmentation avoids the need for dimensionality reduction while preventing multiple collinearity issues.
Solution Approach 2:
The patent extracts and isolates independent features from the original feature set using statistical independence tests. By taking out only the independent features for modeling, it eliminates the multiple collinearity problem while preserving the original meanings of these features, avoiding the information loss that would occur with dimensionality reduction.
2Productivity
If stream data is used for learning and modeling, then real-time user churn analysis is enabled, but existing methods cannot handle such data and require static samples
Solution Approach 1:
The patent applies dynamic analysis methods that can process stream data in real-time. Instead of requiring static samples, it uses statistical tests and independence analysis that can be performed on flowing data, enabling real-time churn prediction while adapting to changing user behavior patterns.
Solution Approach 2:
The patent replaces traditional static sampling mechanisms with statistical independence testing mechanisms. This substitution allows the system to handle stream data by continuously identifying independent features through statistical tests rather than relying on pre-collected static samples.
3Reliability
If churn related features are required to be independent, then model estimation errors are reduced, but the analysis cannot capture complex interactions between features
Solution Approach 1:
The patent segments features into independent and dependent categories using statistical tests. By focusing only on independent features for modeling, it ensures model reliability while handling complexity through systematic feature classification rather than attempting to model all interactions.
Solution Approach 2:
The patent introduces statistical independence tests as an intermediary step between feature collection and modeling. This intermediary process filters and selects features based on their independence properties, simplifying the modeling process while maintaining accuracy without requiring complex interaction analysis.
Data Source
AI summary
A method for mining a churn factor causing user churn for a network application includes: calculating, according to a data universe of churned users, a proportion of a quantity of churned users under each user operation scenario where user churn occurs for a network application in a total quantity of the churned users, and determining multiple user operation scenarios corresponding to multiple proportions sequentially placed in foremost positions in a list of all calculated proportions of user operation scenarios ranked in a descending order; determining churn factors of the multiple user operation scenarios; determining, according to the proportions of the churned users under the multiple user operation scenarios in all the churned users, influence weight values of the churn factors; and determining, when the influence weight value of a churn factor is greater than or equal to the threshold, that the churn factor is a major churn factor.