ML Decision Tree for Proactive KPI Alerting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional alerting systems in computer networks are reactive and rely on hardcoded rules, making them inefficient in managing complex streaming data environments where not all underlying system metrics are available at all times, and they struggle to predict KPI states accurately.
Innovation Solution
A device aggregates key performance indicators (KPIs) and constructs a machine learning-based decision tree to predict KPI states from live performance metric values, enabling proactive and probabilistic alert generation even with partial data availability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional reactive alerting with hardcoded rules is used, then alerting simplicity is maintained, but alerting accuracy and responsiveness deteriorate in complex streaming data environments
Solution Approach 1:
The patent replaces traditional mechanical rule-based alerting systems with a machine learning-based predictive model. The system uses ML algorithms to analyze streaming data and predict future KPI states, substituting the rigid hardcoded rules with adaptive intelligent models that can handle complex patterns in streaming data.
Solution Approach 2:
The patent transforms the alerting approach by changing from static threshold parameters to dynamic predictive parameters. Instead of fixed hardcoded rules, the system uses learned parameters from training data to predict KPI states, allowing the alerting system to adapt to changing conditions in the streaming data environment.
2Adaptability or versatility
If hardcoded alerting rules are used, then system simplicity is maintained, but adaptability to complex streaming data environments deteriorates
Solution Approach 1:
The patent replaces the mechanical rule-based system with a machine learning-based system that can adapt to complex patterns in streaming data. The ML model learns from training data and automatically adjusts its predictions, providing high adaptability without requiring manual rule creation for each scenario.
Solution Approach 2:
The system performs self-learning and self-adjustment through the machine learning model. The model automatically adapts to new patterns in the streaming data through continuous training and prediction, eliminating the need for manual adaptation of alerting rules for each new condition.
3Loss of time
If reactive alerting is used, then implementation simplicity is maintained, but responsiveness and timeliness deteriorate
Solution Approach 1:
The patent implements preliminary action by using predictive models to forecast future KPI states before actual failures occur. The system analyzes streaming data in advance and generates alerts proactively, allowing corrective actions to be taken before problems manifest, thus reducing response time and loss of time.
Solution Approach 2:
The patent replaces reactive mechanical rule-triggered alerts with proactive predictive alerts generated by machine learning models. This substitution enables the system to anticipate future states and generate alerts before conditions deteriorate, significantly improving response time and timeliness.
4Loss of information
If all underlying system metrics are monitored, then measurement completeness is improved, but data processing complexity and dimensionality increase
Solution Approach 1:
The patent extracts and selects only the most relevant metrics for prediction from the full set of available system metrics. The machine learning model identifies and focuses on the critical features that matter for predicting KPI states, filtering out redundant or less informative metrics to reduce processing complexity while maintaining information completeness.
Solution Approach 2:
The patent replaces comprehensive manual monitoring of all metrics with an automated machine learning system that intelligently selects and processes only the necessary metrics. The ML model handles the complexity of metric selection and processing, reducing the burden on the system to process all available data while maintaining accurate predictions.
Data Source
AI summary
In one embodiment, a device in a network aggregates values for a set of key performance indicators (KPIs) for a system the network to form a plurality of KPI states. The device associates a plurality of observed performance metric values from the system with the KPI states. The device constructs a machine learning-based decision tree. Internal vertices of the decision tree represent conditions for the plurality of observed performance metric values and leaves of the tree represent the KPI states. The device predicts a KPI state by using the machine learning-based decision tree to analyze live performance metric values streamed from the system. The device generates a proactive alert based on the predicted KPI state.