Contact Center Workload Forecasting Across Granularities and Horizons
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional timeseries forecasting solutions are inadequate for predicting complex phenomena in contact center metrics, leading to inefficiencies in resource management and inaccurate workload forecasting.
Innovation Solution
A method for selecting forecasting models that includes a first selection process for lower-granularity timeseries, followed by long-term and short-term cross-validation processes to calculate accuracy scores, and identifying a second model for distributing workload across higher-granularity timeseries, enabling improved forecasting for contact centers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional timeseries forecasting solutions are used, then implementation is simple, but forecasting accuracy for complex contact center metrics is inadequate
Solution Approach 1:
The patent segments the model selection process into distinct phases: long-term cross-validation for baseline accuracy and short-term cross-validation for operational accuracy. This segmentation allows comprehensive evaluation without overwhelming complexity, addressing the contradiction between accuracy and complexity by breaking down the selection process into manageable segments.
Solution Approach 2:
The patent performs preliminary long-term cross-validation before short-term cross-validation to pre-identify promising models. This preliminary action filters out inadequate models early, reducing the complexity of subsequent selection while maintaining high forecasting accuracy through multi-stage validation.
2Adaptability or versatility
If single-granularity forecasting is used, then model execution is simple, but it cannot provide forecasts for varying timeseries granularities and operating horizons
Solution Approach 1:
The patent implements a universal forecasting system that can operate at multiple granularities (hourly, daily, weekly) and time horizons (short-term, long-term) using the same model selection framework. The system achieves versatility through a unified cross-validation approach that evaluates models across different temporal scales, eliminating the need for separate forecasting systems for each granularity.
Solution Approach 2:
The patent adds the dimension of time horizon to the forecasting framework by implementing separate long-term and short-term cross-validation processes. This dimensional expansion allows the system to handle varying operating horizons and granularities simultaneously, transforming a single-granularity system into a multi-dimensional forecasting capability.
3Measurement precision
If comprehensive cross-validation is performed, then model accuracy is improved, but computational time and resources increase
Solution Approach 1:
The patent segments comprehensive cross-validation into two distinct phases: long-term cross-validation using historical data and short-term cross-validation using recent data. This segmentation enables thorough model evaluation across different time scales while managing computational resources efficiently by processing different data subsets in separate stages.
Solution Approach 2:
The patent performs long-term cross-validation as a preliminary step to identify promising models before conducting more computationally intensive short-term cross-validation. This preliminary filtering reduces the number of models requiring full validation, thereby reducing overall validation time while maintaining comprehensive accuracy assessment.
Data Source
AI summary
A method for selecting forecasting models for generating timeseries workload forecasts for a contact center covering varying timeseries granularities and operating horizons. The method includes selecting, via a first selection process, a first select forecasting model from first candidate forecasting models for forecasting a workload level in accordance with a lower-granularity timeseries. The first selection process may include the steps of: receiving a first timeseries dataset; defining different timeseries datasets within the first timeseries dataset, including a first shorterm dataset and first longterm dataset; testing, using the first longterm dataset, each first candidate forecasting model in accordance with a first longterm cross-validation process; testing, using the first shorterm dataset, each first candidate forecasting model in accordance with a first shorterm cross-validation process; and calculating, for each of the first candidate forecasting models, an combined accuracy score based on the testing.