Automated Root Cause Identification for Multi-Dimensional Metric Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Determining root causes for metric changes across multiple data dimensions is complicated due to the multi-dimensionality of contributing causes, making it challenging to identify key drivers in data analytics for business intelligence applications.
Innovation Solution
A method and system that retrieve hierarchical time series data from data marts, analyze key-value pairs to produce top driver key-value pairs, and query these pairs based on conditions and a contribution threshold to identify top drivers associated with a metric, such as operational costs or resource usage, by aggregating and joining data across dimensions like location, products, and time periods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual analysis of multi-dimensional data is performed to identify root causes, then measurement precision of root cause identification is improved, but loss of time increases and productivity decreases
Solution Approach 1:
The patent segments the complex multi-dimensional data analysis into distinct hierarchical layers (e.g., regional level, product category level, individual product level). Each layer is analyzed separately to identify contributors at that level, then results are aggregated upward. This segmentation allows automated processing of each layer while maintaining comprehensive root cause identification across all dimensions.
Solution Approach 2:
The patent introduces an intermediary computational system that acts as a mediator between the raw multi-dimensional data and the final root cause identification. This intermediary automatically performs the complex calculations, data aggregation, and contributor identification across multiple dimensions, replacing manual analysis while preserving measurement precision.
2Measurement precision
If comprehensive multi-dimensional data is analyzed to identify all contributing causes, then measurement precision of root cause identification is improved, but device complexity increases
Solution Approach 1:
The patent divides the complex multi-dimensional analysis system into modular components that handle specific dimensions or hierarchical levels independently. Each module processes a subset of dimensions (e.g., one module handles geographic dimensions, another handles product dimensions), reducing the complexity any single component must manage while maintaining comprehensive analysis through integration of all modules.
Solution Approach 2:
The patent transforms the complex multi-dimensional analysis problem by introducing a hierarchical dimension. Instead of analyzing all dimensions simultaneously at the same level, the system organizes analysis in hierarchical layers (e.g., aggregate level → detailed level), adding a temporal or organizational dimension to the analysis process that simplifies the computational complexity.
3Productivity
If automated methods are used to identify top drivers, then productivity increases and loss of time decreases, but measurement precision of root cause identification deteriorates
Solution Approach 1:
The patent performs preliminary actions by pre-processing and organizing multi-dimensional data into standardized hierarchical structures before the actual root cause identification. Data is pre-aggregated, pre-filtered, and pre-structured according to the hierarchical framework, so that the automated identification process works with clean, organized data, maintaining precision while achieving high productivity.
Solution Approach 2:
The patent incorporates feedback mechanisms where the automated system's initial results are validated against expected patterns or thresholds, and adjustments are made iteratively. The system provides feedback loops that allow refinement of automated identification results, ensuring measurement precision is maintained while benefiting from automated processing speed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An aspect of the disclosed technology is a scalable method to derive drivers of change for composite metrics (e.g., cost metrics and ratio metrics) in a time series data set. The disclosed technology comprises an automated mechanism that enables identification and deciphering of one or more drivers, e.g., the largest contributors, or a composite metric change in a scalable manner.