Category: Aggregation
APPROX_COUNT_DISTINCT
APPROX_COUNT_DISTINCT counts the distinct values of a column approximately and is faster than COUNT(DISTINCT) on large data.
Syntax
APPROX_COUNT_DISTINCT(column_name)AMC specifics
- Null values are not counted. The price of the speed is a small error in the result.
- For reach over user_id, COUNT(DISTINCT user_id) remains the documented form: for columns with the very high threshold Amazon names only COUNT and COUNT DISTINCT as permitted aggregations. Whether APPROX_COUNT_DISTINCT is allowed on user_id is not stated there.
Example
Approximate number of distinct creatives per campaign.
SELECT
campaign,
APPROX_COUNT_DISTINCT(creative_id) AS approx_creatives
FROM
dsp_impressions
GROUP BY
1 Checked against the AMC schema, not executed on an AMC instance.