Contact Center Benchmarking Yule-Simpson Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Benchmarking processes in contact centers are subject to the Yule-Simpson effect, leading to misleading assessments of performance differences between contact assignment algorithms due to aggregation of distinct data cross-sections, which can reverse the apparent performance of algorithms that consistently outperform others.
Innovation Solution
A method and system that determine results for multiple contact-agent interactions using different pairing strategies, with a computer processor configured to correct for the Yule-Simpson effect by applying correction factors to ensure accurate relative performance assessment across various partitions such as time periods, agent skills, contact center sites, and benchmarking schedules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If contact center administrators alternate between two algorithms and aggregate performance data over time to compare them, then they can examine performance differences, but the aggregation can result in misleading assessments due to the Yule-Simpson effect that reverse the apparent performance of algorithms
Solution Approach 1:
The patent segments the aggregated performance data into distinct time periods or cohorts to prevent the Yule-Simpson effect. Instead of aggregating all data into a single metric, the system divides the data into separate segments that can be analyzed individually, then combines them using a correction factor that accounts for the segmentations. This segmentation approach preserves the true performance characteristics of each algorithm while still enabling comparative analysis.
2Reliability
If one algorithm consistently outperforms another in each period, then the superior algorithm should be identified, but aggregation can reverse the apparent performance and mischaracterize which algorithm is better
Solution Approach 1:
The patent implements a feedback mechanism that calculates a correction factor based on the segmentation of data. This correction factor is then applied to the aggregated performance metrics to adjust for the Yule-Simpson effect. The feedback loop ensures that the final performance assessment reflects the true relative performance of algorithms by incorporating information about how data was segmented and distributed across different periods.
3Ease of operation
If administrators use simple aggregation of performance data, then the analysis process is straightforward, but the results can be misleading due to unequal distribution of contacts across algorithms
Solution Approach 1:
The patent changes the parameters used in performance aggregation by introducing a correction factor that adjusts for unequal contact distribution. Instead of using a simple weighted average of performance metrics, the system modifies the aggregation formula to incorporate the correction factor, which is derived from the distribution characteristics of contacts across different algorithms and time periods. This parameter change maintains computational simplicity while significantly improving measurement accuracy.
Data Source
AI summary
Techniques for benchmarking pairing strategies in a contact center system are disclosed. In one particular embodiment, the techniques may be realized as a method for benchmarking pairing strategies in a contact center system including determining results for a first plurality of contact-agent interactions, determining results for a second plurality of contact-agent interactions, and determining combined results across the first and second pluralities of contact-agent interactions corrected for a Yule-Simpson effect.