Analytics context

Batch analytics in FeatureQL enables large-scale feature computation across entity populations. This section covers the tools for connecting to external data sources and processing features at scale.

Key capabilities

Data source integration:

• EXTERNAL_COLUMNS(): Map features to table columns declaratively
• EXTERNAL_SQL(): Map features to custom SQL queries for complex preprocessing
• BIND_KEYSET(): Define entity populations as inputs to your FeatureQL queries

Analytics operations:

• Grouping & aggregation: GROUP BY with standard SQL aggregation functions
• Unnesting: UNNEST() arrays for rows as a table for final presentation
• Filtering & sorting: WHERE, ORDER BY, LIMIT, OFFSET clauses

Hybrid queries:

• Embed FeatureQL within SQL for seamless integration with existing analytics tools
• Use FeatureQL as CTEs or subqueries in standard SQL workflows

Query structure

WITH
    [feature definitions]
SELECT
    [features to compute]
FOR
    [bind entity keys]
WHERE [filter]
GROUP BY [aggregate]
ORDER BY [sort]
LIMIT [rows]

Note: Analytics clauses (WHERE, GROUP BY, ORDER BY, LIMIT) work only in evaluation queries, not in CREATE FEATURES statements.