APPROX_COUNTD_AGG

When the calculation data is very large, the traditional accurate deduplication count may not be calculated, and the approximate deduplication count can be used to quickly calculate the result.

• Only supports use in real-time data.

• This function only supports the following database systems: CLICKHOUSE, ORACLE (12c r1, 12.1.0.2 and above), SQLSERVER (2019 and above), PRESTO, REDSHIFT, VERTICA. When the database system is REDSHIFT, due to the limitation of REDSHIFT, only aggregation operations that need to be sorted (such as deduplication count, median, percentile, approximate deduplication count) can be performed on the same field at the same time. And this aggregation operation on 2 or more fields at the same time will cause errors. When the database system is VERTICA, due to the limitation of VERTICA, only one of (precise) deduplication counting or approximate deduplication counting can be used at the same time.

• Supports the use of a parameter of any type.

When the user's horizontal axis is the dimension field "Day", the value returned by the calculation indicator APPROX_COUNTD_AGG() on the vertical axis is the approximate deduplication number of daily sales.

When the user's horizontal axis is the dimension field "Month", the value returned by APPROX_COUNTD_AGG() is the approximate deduplication number of monthly sales.

For example, calculate the approximate daily deduplication count for "ORDERID", as shown in the following figure: