Skip to main content
Version: 2.5

percentile_union

Description​

Aggregates PERCENTILE data.

Syntax​

percentile_union(expr);

Parameters​

expr: The supported data type is PERCENTILE.

Return value​

Returns a PERCENTILE value.

Examples​

Example 1: Use percentile data in materialized views.

Create a table.

CREATE TABLE sales_records(
    record_id int, 
    seller_id int, 
    store_id int, 
    sale_date date, 
    sale_amt bigint
) distributed BY hash(record_id) 
properties("replication_num" = "3");

Create a materialized view based on the sale_amt column of the table.

create materialized view mv as
select store_id, percentile_union(percentile_hash(sale_amt))
from sales_records
group by store_id;

Example 2: Load PERCENTILE data.

Create an aggregate table that contains a PERCENTILE column sale_amt_per.

CREATE TABLE sales_records(
    record_id int, 
    seller_id int, 
    store_id int, 
    sale_amt_per percentile percentile_union
) ENGINE=OLAP
AGGREGATE KEY(`record_id`, `seller_id`, `store_id`)
COMMENT "OLAP"
DISTRIBUTED BY HASH(`record_id`) BUCKETS 3
PROPERTIES (
    "replication_num" = "3",
    "storage_format" = "DEFAULT"
);

Query data from sale_amt_per.

select percentile_approx_raw(percentile_union(sale_amt_per), 0.99) from sales_records;

Load data that contains PERCENTILE data into the sales_records table.

curl --location-trusted -u root
    -H "columns: record_id, seller_id, store_id,tmp, sale_amt_per =percentile_hash(tmp)"
    -H "column_separator:,"
    -T a http://<ip:port>/api/test/sales_records/_stream_load