Shifted Time-Series Correlation for Root Cause Event Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods are inefficient in quickly and effectively identifying candidate time series that have a high probability of causing an event in a reference time series, particularly in cloud computing environments where numerous connections and data sources need to be analyzed.
Innovation Solution
A computer-implemented method involving correlation coefficient calculations, differentiation, shifting, and smoothing of time series data to identify candidate time series with a high likelihood of causing an event, using techniques such as Pearson correlation, Savitzky-Golay filtering, and simplified time series construction based on event detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional correlation analysis methods are used to identify root cause time series, then the analysis can be performed, but the computational complexity increases significantly and the analysis time becomes too long when dealing with numerous candidate time series in cloud computing environments
Solution Approach 1:
The patent segments the complex correlation analysis process into distinct phases: preprocessing (smoothing, differentiation), correlation coefficient calculation, and result filtering. This segmentation allows each phase to be optimized independently, reducing overall computational complexity while maintaining analysis accuracy.
Solution Approach 2:
The patent applies preliminary actions by performing smoothing and differentiation operations on time series data before calculating correlation coefficients. These preprocessing steps simplify the data structure and reduce noise, making subsequent correlation calculations more efficient and accurate.
2Measurement precision
If multiple transformations (differentiation, shifting, smoothing) are applied to time series to improve correlation accuracy, then the precision of root cause identification improves, but the computational time and complexity increase
Solution Approach 1:
The patent implements partial action by selectively applying transformations only when necessary. The system calculates correlation coefficients with different transformation levels and selects the most appropriate one, avoiding unnecessary computational overhead while maintaining sufficient accuracy for root cause identification.
Solution Approach 2:
The patent changes parameters by applying different smoothing windows, differentiation orders, and shifting amounts to optimize the correlation analysis. By adjusting these parameters systematically, the method achieves high precision in identifying root cause time series while managing computational requirements.
3Reliability
If all candidate time series are analyzed in detail to ensure accurate root cause identification, then the reliability of the analysis improves, but the time required to process numerous candidate series becomes prohibitive
Solution Approach 1:
The patent applies local quality by focusing computational resources on the most promising candidate time series. After initial correlation analysis, the system identifies and performs detailed analysis only on candidates with correlation coefficients above a certain threshold, ensuring high reliability for the most likely root causes while maintaining high throughput.
Solution Approach 2:
The patent implements skipping by rapidly filtering out unlikely candidates through quick correlation coefficient calculations. The system processes through large numbers of candidates efficiently, skipping detailed analysis for those with low correlation values and concentrating resources on promising candidates.
Data Source
AI summary
The disclosure concerns the time-based correlation of time series for the root cause analysis of application monitoring data, observability data, and data observability data in the field of application monitoring and observability. The object of the disclosure is to find at least one candidate time series that has a high chance/probability for causing an event in a reference time series. The method shall be time-based and not frequency based. The method includes: constructing a simplified reference time series retaining changes in the reference time series and setting other points to zero; constructing a simplified candidate time series retaining changes in the candidate time series and setting other points to zero; calculating a similarity metric between the simplified candidate time series and the simplified reference time series; and reporting an occurrence of a computing event in response to the similarity metric exceeding a threshold value.