Data Versioning Engine for Multi-Scenario Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data warehouse systems are inadequate for handling different data versions for planning applications, as they primarily focus on backward-looking reporting rather than forward-looking forecasting, making it difficult to extend reporting data to reflect various options in a consistent manner.
Innovation Solution
Implementing data and metadata versioning through a versioning engine that creates additional data and metadata variants based on user input, allowing visualization of actual data evolution into plan data for forecasting purposes and enabling comparison of plans on the same metadata.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If data warehouse systems use a single truth model for reporting, then data consistency for historical reporting is maintained, but the system cannot support multiple planning scenarios and forecasting options
Solution Approach 1:
The patent segments data into multiple versions (actual data version and plan data versions) where each version represents a different scenario. The versioning engine creates separate data variants for different planning scenarios while maintaining the original actual data intact, allowing multiple scenarios to coexist without compromising the stability of the source data.
Solution Approach 2:
The patent adds a versioning dimension to the data model, transforming a single-truth data structure into a multi-version data structure. Each data element now has an additional attribute (version identifier) that distinguishes between actual data and various plan data scenarios, enabling the system to handle multiple planning options while maintaining data consistency through the versioning framework.
2Loss of information
If data warehouse systems store only actual historical data, then data integrity is preserved, but the system cannot visualize evolution into plan data for forecasting
Solution Approach 1:
The versioning engine performs preliminary actions by creating plan data variants from actual data before the planning decisions are finalized. This allows the system to prepare multiple forecasting scenarios in advance, visualizing how actual data might evolve under different planning assumptions, while preserving the original actual data for reference and comparison.
3Adaptability or versatility
If the system creates multiple data variants for planning, then forecasting flexibility is improved, but the complexity of managing data versions increases
Solution Approach 1:
The versioning engine acts as an intermediary between the actual data and the multiple plan data scenarios. It automatically manages the creation, storage, and retrieval of data variants, handling the complexity of version control, relationships, and consistency. This intermediary layer shields users from the underlying complexity while providing flexible forecasting capabilities through a simplified interface.
4Adaptability or versatility
If data warehouse systems maintain single metadata structure, then system simplicity is maintained, but the system cannot reflect changes to original corporate metadata for new product lines
Solution Approach 1:
The patent segments metadata into version-specific metadata structures, where each data variant has its corresponding metadata version. This allows the system to maintain the original corporate metadata structure intact while creating new metadata variants that reflect changes for new product lines or scenarios, enabling metadata adaptability without compromising system simplicity through uncontrolled complexity.
Data Source
AI summary
Embodiments implement data and meta data versioning in order to adapt reported data (“actuals”) for planning purposes. A versioning engine receives from an operative system (e.g. ERP system), root variants. These root variants may comprise existing actual data and corresponding underlying corporate meta data. Based upon user input, the versioning engine creates from these root variants, additional variants of the data and/or meta data. A new data variant may be based upon the original corporate meta data unchanged (e.g. for projecting existing product lines). A data variant may alternatively be based upon a new meta data variant reflecting changes to the original corporate meta data (e.g. for adopting an entirely new product line). By effectively depicting relationships between actual data and plan data, and between that data and its underlying meta data, versioning allows a user to visualize evolution of actual data into plan data for forecasting purposes.