Child Datasets

Child datasets are specialized views derived from existing datasets that allow you to create focused versions of your data. They enable you to segment your datasets for different workflows, apply targeted analysis, and take different actions on different subsets of your information.

What Are Child Datasets?

Think of a child dataset as a filtered version of an existing dataset that inherits all the parent's data and columns but can apply its own constraints and transformations. Child datasets maintain a live connection to their parent, automatically receiving updates as the parent data changes.

The key insight is that child datasets allow you to create multiple specialized views of the same underlying data, each optimized for different business purposes or workflows.

How Child Datasets Work

Live Inheritance

Child datasets automatically inherit all columns from their parent dataset. This includes both the original data columns and any agent, formula, tool, or taxonomy columns that exist on the parent.

Continuous Updates

The connection between parent and child is live - when new columns are added to the parent dataset, they automatically appear in all child datasets. When parent data updates, children see those changes immediately.

Restricted Editing

While child datasets can see and use all parent columns, they can only edit columns they own. The parent dataset maintains control over its original columns, ensuring data integrity across the family of related datasets.

Independent Column Creation

Child datasets can create their own new columns - agents, formulas, tools, or taxonomy columns that are specific to their filtered view of the data. These child-specific columns don't affect the parent or sibling datasets.

Common Use Cases

Workflow Segmentation

Create different child datasets for different stages of your business process. For example, from a customer feedback dataset, you might create separate children for:

• Positive feedback requiring thank-you responses
• Negative feedback needing immediate attention
• Neutral feedback for routine processing

Each child can have specialized agents and workflows appropriate for its segment.

Primary Key Transformation

Child datasets can apply filters that effectively change the primary key structure, which directly affects how triggers work for that dataset. Your parent dataset might use a simple ID as its primary key, but a child dataset could filter for specific conditions and use a combination of multiple fields as its effective primary key.

This transformation is particularly powerful for controlling when columns execute. By changing the primary key structure through filtering, you can create child datasets where triggers fire based on different data combinations or timing patterns than the parent dataset.

Parallel Processing

Split a single dataset into multiple children and apply completely different analysis approaches to each. This allows you to compare different methodologies or run multiple experiments simultaneously on the same base data.

Specialized Analysis

Create child datasets focused on specific business questions. From an orders dataset, you might create children for:

• High-value customer analysis
• Geographic performance tracking
• Seasonal trend identification

Each child can have specialized columns and agents tailored to its analytical focus.

Practical Benefits

Organizational Clarity

Child datasets help organize your work by creating clear, purpose-driven views of your data rather than trying to handle all scenarios in a single complex dataset.

Workflow Isolation

Different teams or processes can work with their relevant child datasets without interfering with each other's columns or analysis approaches.

Experimentation Safety

You can test new analysis approaches or agent configurations on child datasets without risking the stability of your main data processing workflows.

Scalable Architecture

As your data analysis needs grow, child datasets provide a clean way to scale by creating specialized views rather than overcomplicating parent datasets.

Relationship Dynamics

Parent Authority

The parent dataset controls the foundational data structure and original columns. Changes to parent columns affect all children, maintaining consistency across the data family.

Child Autonomy

Children have complete control over their own columns and can create specialized processing that doesn't affect other parts of the system.

Sibling Independence

Child datasets don't directly affect each other - they share a common parent but operate independently in their specialized domains.

Live Synchronization

All relationships stay current automatically. Parent updates flow to children immediately, ensuring the entire family of datasets stays synchronized with the latest data.

Design Considerations

Purpose-Driven Creation

Create child datasets with specific business purposes in mind. Each child should have a clear reason for existing and a defined scope of responsibility.

Filter Strategy

Design child dataset filters to create meaningful business segments. Consider how the filtered data will be used and what specialized processing it needs.

Column Ownership Planning

Think about which columns belong on the parent (shared across all use cases) versus which should be specific to individual children (specialized processing).

Primary Key Implications

Understand how your child dataset filters might affect the effective primary key structure and plan your triggers and processing accordingly.

Advanced Customization

For users comfortable with EraQL, Cotera's query language, child datasets can be created with completely custom logic rather than simple filters. This advanced option allows you to write arbitrary EraQL transformations to reshape, aggregate, or transform the parent data in sophisticated ways.

This capability transforms child datasets from simple filtered views into powerful data transformation tools, though it requires deeper technical knowledge to implement effectively.