Skip to main content
Version: 0.0.36

Hierarchies

The Hierarchies tab in Custom Models plays a pivotal role in structuring data analysis by forming hierarchical relationships based on data sources and their semantic attributes. The system identifies cross-hierarchies across multiple data sources during semantic layer operations and presents them here for further exploration.

Understanding Hierarchies​

Hierarchies are integral to the semantic layer, offering a robust framework for data analysis. Internal processes automatically generate these hierarchies, enabling implicit accessibility to users. When properly configured, hierarchies enhance the model's analytical capabilities by structuring data logically.

Key Features​

  • Cross-Hierarchy Recognition: Automated identification of hierarchical structures across connected data sources, showcasing them in the Hierarchies tab.
  • Semantic Inheritance: Hierarchies inherit attributes from semantic data sources, facilitating coherent data analysis.

Hierarchy Management​

While system-generated hierarchies are crucial for analysis, users can intervene in several ways:

Deactivation​

  • Purpose: Disable existing or automatically generated hierarchies if they lack relevance or do not align with business or analytical needs.
  • Considerations: Be cautious and ensure the hierarchical assumptions made by the language model are sound before deactivating.

Adding Hierarchies​

  • User Contribution: Users or modelers can introduce additional hierarchies and link attribute associations to refine analytical structures.
  • Locking Mechanisms: Protect valuable hierarchies by locking them during the modeling phase to prevent alteration through training processes.

Best Practices​

  • Preserve Initially Generated Hierarchies: Retain initial hierarchies unless clear reasons for modification arise.
  • Evaluate Hierarchy Validity: Ensure new or existing hierarchies logically support analytical objectives before implementing changes.
  • Lock Essential Structures: Utilize locks to safeguard hierarchies deemed essential for maintaining analytical integrity.

Conclusion​

The Hierarchies feature in Lakehousecat Custom Models provides a structured approach to data interaction, enhancing analytical precision through semantic logic. By effectively managing hierarchies and understanding their role, users can drive impactful data analysis and foster informed decision-making within their organizations.