Forecast Model Selection Graph for Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing forecasting methods often rely on single predictive models, which may not capture the full complexity of a physical process, leading to suboptimal accuracy and variability in predictions.
Innovation Solution
A computer-implemented system and method for combining forecasts from multiple models using a forecast model selection graph, where nodes are arranged in parent-child relationships, allowing for the transformation and selection of node forecasts into a combined forecast based on selection criteria, such as performance and metadata analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple forecast models are combined, then forecast accuracy is improved, but computational complexity and system complexity increase
Solution Approach 1:
The forecast model selection graph is divided into hierarchical levels with multiple nodes at each level. Each node represents a specific forecast model or combination strategy, allowing the system to segment the complex combination process into manageable, discrete decision points that can be evaluated independently.
Solution Approach 2:
The patent introduces a hierarchical dimension to the model selection process by organizing nodes in parent-child relationships across multiple levels. This dimensional structure transforms the flat complexity of model combination into a structured, multi-level evaluation framework that systematically reduces complexity at each stage.
2Reliability
If multiple forecast models are combined, then forecast reliability is improved, but processing time and computational resources increase
Solution Approach 1:
The system pre-establishes the forecast model selection graph structure with defined parent-child relationships and evaluation criteria before actual forecasting occurs. Model performance metrics and selection rules are predetermined, allowing the system to quickly traverse the hierarchy during forecasting without performing complex real-time analysis.
Solution Approach 2:
The forecast model selection graph allows for dynamic adaptation where the system can adjust which nodes are evaluated based on data characteristics and performance history. The hierarchical structure enables flexible pruning of less relevant model combinations, reducing processing time while maintaining reliability through adaptive model selection.
Data Source
AI summary
Systems and methods are provided for evaluating performance of forecasting models. A plurality of forecasting models may be generated using a set of in-sample data. Two or more forecasting models from the plurality of forecasting models may be selected for use in generating a combined forecast. An ex-ante combined forecast may be generated for an out-of-sample period using the selected two or more forecasting models. The ex-ante combined forecast may then be compared with a set of actual out-of-sample data to evaluate performance of the combined forecast.