Monotonic Binning Rule Generation for Computer Operation Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning approaches for predicting characteristics of computer operations are limited in handling different types of variables, fail to leverage user domain knowledge, and do not provide mechanisms for users to modify or remove constraints during the rule generation process, leading to suboptimal performance in detecting undesirable computer operations.
Innovation Solution
A system that includes a rule generation system capable of preprocessing historical data, performing monotonic binning, and optimizing cutoffs to generate rules that can handle various types of variables and incorporate user domain knowledge, allowing users to define characteristics, select variables, and adjust rule generation parameters to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional machine learning approaches are used for predicting characteristics of computer operations, then the system can identify trends in data, but the system fails to handle different types of variables effectively and cannot leverage user domain knowledge
Solution Approach 1:
The patent segments the rule generation process into distinct modules: data preprocessing module, monotonic binning module, rule generation module, and optimization module. Each module handles specific aspects of variable processing, allowing the system to effectively manage different variable types while maintaining prediction reliability through specialized processing at each stage.
Solution Approach 2:
The patent introduces monotonic binning as an intermediary transformation step that converts various variable types into a unified monotonic scale. This intermediary process enables the system to handle diverse variable types (categorical, numerical, temporal) while preserving the relationship between variables and target characteristics, thereby maintaining prediction accuracy.
2Ease of operation
If conventional machine learning approaches are used, then models can be trained from existing data, but users cannot modify or remove constraints during the rule generation process
Solution Approach 1:
The patent implements a dynamic rule generation process where users can iteratively adjust parameters including the target variable, variable selection criteria, binning configurations, and constraint conditions. The system re-executes the rule generation pipeline with each parameter change, allowing users to optimize prediction performance while maintaining ease of operation through an interactive interface.
3Measurement precision
If monotonic binning is performed on selected variables, then the system can create bins with monotonically increasing or decreasing prevalence, but the process requires determining optimal cutoffs which increases system complexity
Solution Approach 1:
The patent implements an optimization module that uses feedback from prediction performance metrics to automatically adjust cutoff values for each variable's bins. The system evaluates different cutoff configurations based on their ability to correctly predict target characteristics and iteratively refines the cutoffs to maximize prediction accuracy, thereby achieving high measurement precision without requiring manual intervention.
Data Source
AI summary
Techniques are disclosed that relate to generating a rule for performing predictions of a characteristic of computer operations. A computer system may receive historical data that describes executed computer operations, including variables associated with those executed computer operations, and user input specifying desired properties of the performed predictions. The computer system determines, for a given variable, bins having ranges specified using the variable. The bins may be formed such that, when the executed computer operations are grouped into the bins, a prevalence of the characteristic monotonically increases or decreases from bin to bin across a bin ordering that is based on the ranges. The computer system then determines one or more cutoffs for one or more of the variables based on the desired properties from user's inputs and the determined bins. The computer system generates the rule based on the one or more cutoffs and the one or more variables.