Skip to content

Aggregates

You will learn aggregate functions, intervals, compare, and URI normalization.

Functions

UI label API aggregate Needs numeric field?
Count count No
Sum sum Yes
Average avg Yes
Max max Yes
P50 p50 Yes
P95 p95 Yes
P99 p99 Yes
Search (events) events Projects learned fields

Time buckets (interval)

UI label Token
None — table only (omit)
1 minute 1m
5 minutes 5m
15 minutes 15m
1 hour 1h
1 day 1d

With an interval, buckets are zero-filled across the selected range for charts. Numeric aggregate rows include sample_count, the number of non-null field values that participated in the calculation. Empty, zero-filled buckets report zero samples.

In Query, select Inspect contributing events on a result row to search that bucket without losing the aggregate. Percentile and maximum investigations open the upper metric tail; minimum opens the lower tail. Select Back to aggregate to restore the original result.

A percentile is a distribution value rather than an event identity, so the inspection shows candidate tail events rather than claiming one event caused the percentile.

Normalize URI paths

When Group by is set, enable Normalize URI paths (normalize_uri: true) to collapse UUID/hex/numeric path segments to {id}.

Example:

/v1/workspaces/550e8400-e29b-41d4-a716-446655440000/search
→ /v1/workspaces/{id}/search

Ignored for Search / events mode.

Period compare

On Query Aggregate, set Compare to Prior period. The same aggregate runs over the equal-length window immediately before the primary range. Rows include compare_value, delta, and pct_change; charts overlay the prior series. Saving as an aggregate view keeps the compare setting.

REST and MCP use compare=prior_period. Not available with aggregate=events.