Predictive Analytics Delta Thresholding for Metric Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Predictive analytics in computer systems face challenges in accurately identifying entities that require updated predictions based on changing metric values, as existing methods do not effectively account for significant deltas in resource usage over time.
Innovation Solution
A method is introduced where deltas between current and previous metric values are calculated, and an average delta is determined for a subset of entities with significant changes. If an entity's delta exceeds this average and was not previously used in predictions, a new prediction is created, leveraging statistical algorithms to analyze these changes and update predictions accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If predictions are updated for all entities based on metric value changes, then prediction coverage is improved, but computational resources and processing time are wasted on entities with insignificant changes
Solution Approach 1:
The patent applies parameter changes by introducing a threshold parameter (average delta) that dynamically determines which entities should trigger prediction updates. Instead of updating all entities uniformly, the system calculates the average delta across all entities and uses this parameter to filter which entities' predictions should be updated, thereby optimizing the balance between prediction accuracy and processing efficiency
Solution Approach 2:
The patent implements local quality by applying different update strategies to different entities based on their individual delta values. Entities with deltas above the average threshold receive prediction updates, while those below the threshold do not. This localized approach ensures that computational resources are concentrated on entities that truly need updated predictions, improving overall system efficiency
2Reliability
If predictions are updated frequently to capture all metric changes, then prediction freshness is improved, but system overhead and false positives increase
Solution Approach 1:
The patent applies partial action by updating predictions for only a subset of entities rather than all entities. By calculating the average delta and selectively updating only those entities whose metric changes exceed this average, the system achieves sufficient prediction freshness for critical entities while avoiding the overhead of universal updates, thus reducing false positives and system complexity
3Loss of information
If all metric value changes are used to trigger predictions, then comprehensive monitoring is achieved, but noise and false alarms increase
Solution Approach 1:
The patent implements preliminary action by calculating the average delta threshold before making prediction decisions. This pre-computed parameter serves as a filter that distinguishes significant metric changes from noise. By establishing this baseline threshold in advance, the system can comprehensively monitor all entities while reliably filtering out false alarms, maintaining both monitoring completeness and signal quality
Data Source
AI summary
In an embodiment, deltas are calculated between respective current metric values for respective entities and previous metric values for the respective entities. A subset of the deltas is determined. A sum of the subset is calculated, and the sum is divided by a number of the subset to create an average delta for the subset. If one of the respective entities has one of the deltas that is greater than or equal to the average delta for the subset and the one of the respective entities was not previously used to create the previous prediction, then a current prediction is created.