​
Overview

Use the WHERE clause to filter summary or timeseries rows before Horizon AI Query evaluates aggregations or grouping. When a GROUP BY clause is present, the filter runs first so only matching rows contribute to each group. The WHERE clause evaluates SQL-style predicates against the same column expressions that are available in the select and group_by clauses.

​
Supported operators

WHERE predicates support these operators:
  • Comparison: >, >=, <, <=, =, !=
  • Logical: AND, OR, NOT
  • Parentheses for explicit precedence
  • String literals enclosed in single quotes (for example, 'long_sleep')
Comparison operands can reference table columns (for example, stage_rem_second) or Source columns (for example, source_type).
Column identifiers follow the Continuous Query column names. See Column Mappings for a full list.

​
Filter rows before grouping

import vitalx.aggregation as va

va.select(
    va.group_key("*"),
    va.Sleep.score().mean(),
    va.Sleep.col("stage_asleep_second").sum(),
).group_by(
    va.date_trunc(va.Sleep.index(), 1, "day"),
    va.Sleep.col("state"),
).where(
    "type = 'long_sleep'"
)

​
Combine multiple device sources

Filter for specific provider and source type combinations with parentheses to manage precedence. 
va.select(
    va.Sleep.score().mean(),
).where(
  "(source_type = 'watch' AND source_provider = 'apple_health_kit') OR (source_type = 'ring' AND source_provider = 'oura')"
)

​
Filter without grouping

You can apply a WHERE clause to ungrouped queries to trim the dataset before computing aggregates. 
va.select(
    va.Sleep.score().mean()
).where("source_type = 'watch' AND source_provider = 'apple_health_kit'")
In the JSON DSL, supply the same predicate string via the top-level where property. 
{
  "select": [
    { "func": "mean", "arg": { "value_macro": "sleep_score" } }
  ],
  "where": "source_type = 'watch' AND source_provider = 'apple_health_kit'"
}

​
Evaluation order

  1. Apply the WHERE predicate to the incoming rows.
  2. Evaluate the GROUP BY clause (if provided) on the filtered rows.
  3. Calculate the aggregates in the SELECT clause.
This order ensures filters never exclude rows after grouping has started, keeping results consistent with SQL semantics.