Hybrid Forecasting System for Cloud Resource Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network demand forecasting models in cloud computing environments often rely on either cyclical or event-based forecasting, leading to inefficient resource allocation due to the lack of integration of both paradigms and excessive parameter selection, which can result in suboptimal provisioning of resources.
Innovation Solution
The implementation of a hybrid forecasting system that uses evolutionary algorithms for joint hyperparameter selection of cyclical and event-based forecasting models, allowing for the optimization of resource allocation by predicting future demand through a combination of historical and event-based forecasting features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a hybrid forecasting system combining cyclical and event-based models is implemented, then forecasting accuracy is improved, but system complexity increases due to joint hyperparameter selection
Solution Approach 1:
The patent combines cyclical forecasting models and event-based forecasting models into a single hybrid system. The evolutionary algorithm jointly optimizes hyperparameters for both forecasting paradigms, merging previously separate selection processes into one unified approach that improves overall forecasting accuracy while managing complexity through integration.
Solution Approach 2:
The system performs joint hyperparameter selection for multiple forecasting models simultaneously. By changing the parameters of both cyclical and event-based models together through evolutionary optimization, the system achieves better forecasting accuracy than separate model optimization, while the automated parameter tuning manages the complexity of having multiple models.
2Productivity
If evolutionary algorithms are used for joint hyperparameter selection, then resource allocation efficiency is improved, but computational time increases
Solution Approach 1:
The system performs hyperparameter selection and model optimization in advance before actual resource allocation is needed. By pre-computing the optimal forecasting model configuration using evolutionary algorithms, the system prepares the best resource allocation strategy ahead of time, improving real-time allocation efficiency while accepting the computational time cost during the preliminary optimization phase.
Solution Approach 2:
The evolutionary algorithm automatically performs joint hyperparameter selection without requiring manual intervention. The system self-optimizes its forecasting models by autonomously tuning parameters for both cyclical and event-based components, improving resource allocation efficiency through automated optimization while the computational time is invested in this self-service optimization process.
3Measurement precision
If more forecasting features are integrated into the model, then prediction accuracy improves, but model complexity increases
Solution Approach 1:
The patent merges historical forecasting features with event forecasting features into a unified hybrid model. By integrating multiple feature types from different forecasting paradigms (cyclical and event-based) into a single comprehensive model, the system achieves improved prediction accuracy while managing complexity through unified model architecture and joint hyperparameter optimization.
Data Source
AI summary
Approaches for optimizing network demand forecasting models and network topology using hyperparameter selection are provided. An approach includes defining a pool of features that are usable in models that predict demand of network resources, wherein the pool of features includes at least one historical forecasting feature and at least one event forecasting feature. The approach also includes generating, using a computer device, an optimal model using a subset of features selected from the pool of features. The approach further includes predicting future demand on a network using the optimal model. The approach additionally includes allocating resources in the network based on the predicted future demand.