Automatic Root Cause Analysis Using Decision Trees
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current root cause analysis techniques are inefficient for analyzing static data in real-time or near real-time, failing to effectively identify root causes of faults in systems where data is static.
Innovation Solution
A system and method for automatic root cause analysis that generates a decision tree based on current time-window records, selecting features that maximize information gain and using conditional entropy to classify causal and root causes, enabling near-real-time detection with statistical significance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional root cause analysis techniques are used on static data, then analysis can be performed, but the analysis cannot be done in real-time or near real-time
Solution Approach 1:
The patent transforms static data analysis into a dynamic process by implementing a sliding time-window approach that continuously updates the analysis as new data arrives. The decision tree model is rebuilt periodically with updated time windows, enabling the system to adapt to changing conditions while maintaining real-time or near real-time analysis capability.
Solution Approach 2:
The patent performs preliminary actions by pre-processing data into structured formats, pre-defining the decision tree structure with relevant features, and pre-establishing the analysis framework before actual root cause analysis is needed. This preparation enables rapid real-time analysis when faults occur.
2Measurement precision
If a decision tree is generated using all records in a large time window, then statistical significance is improved, but the complexity of processing and generating the decision tree increases
Solution Approach 1:
The patent segments the large time window into smaller, manageable intervals and processes data in a sliding window fashion. Instead of loading all historical data at once, the system divides the time window into segments that can be processed incrementally, reducing memory requirements and computational complexity while maintaining statistical significance through sufficient sample sizes in each window.
Solution Approach 2:
The patent applies partial action by selecting only the most relevant features for the decision tree construction rather than processing all possible attributes. The system identifies and focuses on key features that maximize information gain, performing analysis on a subset of critical data that provides sufficient statistical significance without the overhead of processing complete datasets.
3Loss of information
If manual root cause analysis is performed, then detailed analysis can be conducted, but the process is time-consuming and cannot be automated
Solution Approach 1:
The patent implements self-service by automating the entire root cause analysis process through machine learning models that automatically generate decision trees, select features, and identify root causes without human intervention. The system autonomously processes fault data, performs analysis, and generates results, eliminating the time-consuming manual analysis while preserving detailed information through comprehensive feature evaluation and decision tree generation.
4Speed
If a small time window is used for real-time detection, then response time is improved, but statistical significance is reduced
Solution Approach 1:
The patent dynamically adjusts the time window size based on the specific analysis needs and data characteristics. The system can modify the window parameters to optimize between real-time detection speed and statistical significance, allowing flexible adaptation to different fault types, data frequencies, and analysis requirements without being constrained to a fixed window size.
Data Source
AI summary
A system, method, and computer program product are provided for automatic root cause analysis. In operation, a root cause analysis system identifies at least one event associated with one or more records for which to perform a root cause analysis. The root cause analysis system performs a root cause analysis of the event by automatically generating a decision tree based on all records in the current time-window such that each leaf in the decision tree represents the probabilities for class labels of a target variable and each branch in the decision tree represents a feature that leads to a corresponding class label probability. The root cause analysis system automatically generates the decision tree by automatically selecting at each step the feature that maximizes information gain based on a current subset of data. The root cause analysis system then classifies which conditioned feature is a causal factor and which is a root cause of the event by using a conditional entropy equation on each branch leading to the tree leaf. The root cause analysis is repeatedly performed on sequential time-window sets of records gathered, per a sufficiently small time window for near-real-time root cause detection, yet sufficiently large records set for statistical significance.