startofweek

This page explains how to use the startofweek function in APL.

Use the startofweek function in APL to round a datetime value down to the start of the week. The function returns the preceding Sunday at midnight (00:00:00) for the date that contains the given datetime value. You can optionally shift the result by a specified number of weeks using the offset parameter.

You can use startofweek to bin events into weekly buckets for aggregation, reporting, and trend analysis across log, trace, and security datasets.

Use it when you want to:

  • Group events by week for weekly summaries and dashboards.
  • Align timestamps to week boundaries for consistent aggregation.
  • Compare metrics across different weeks.

Usage

Syntax

startofweek(datetime [, offset])

Parameters

NameTypeDescription
datetimedatetimeThe input datetime value.
offsetlongOptional: The number of weeks to offset from the input datetime. Default is 0.

Returns

A datetime representing the start of the week (Sunday at 00:00:00) for the given date value, shifted by the offset if specified.

Use case examples

Count requests per week to identify weekly traffic patterns.

Query

['sample-http-logs']
| extend week_start = startofweek(_time)
| summarize request_count = count() by week_start
| sort by week_start asc

Run in Playground

Output

week_startrequest_count
2025-01-05T00:00:00Z10234
2025-01-12T00:00:00Z11587
2025-01-19T00:00:00Z9876

This query bins each HTTP request to the start of its week and counts the total requests per week.

Track the weekly average span duration for each service.

Query

['otel-demo-traces']
| extend week_start = startofweek(_time)
| summarize avg_duration = avg(duration) by week_start, ['service.name']
| sort by week_start asc

Run in Playground

Output

week_startservice.nameavg_duration
2025-01-05T00:00:00Zfrontend00:00:01.2340000
2025-01-12T00:00:00Zfrontend00:00:01.1750000
2025-01-19T00:00:00Zfrontend00:00:01.2890000

This query groups trace spans by week and service, then calculates the average span duration for each combination.

Monitor weekly server error trends to detect weeks with unusual failure activity.

Query

['sample-http-logs']
| where toint(status) >= 500
| extend week_start = startofweek(_time)
| summarize error_count = count() by week_start
| sort by week_start asc

Run in Playground

Output

week_starterror_count
2025-01-05T00:00:00Z45
2025-01-12T00:00:00Z67
2025-01-19T00:00:00Z38

This query filters for server errors and counts them per week to reveal weekly error patterns.

  • endofweek: Returns the end of the week for a datetime value.
  • startofday: Returns the start of the day for a datetime value.
  • startofmonth: Returns the start of the month for a datetime value.
  • startofyear: Returns the start of the year for a datetime value.
  • week_of_year: Returns the ISO 8601 week number from a datetime value.

Other query languages