The traditional GROUP BY function organizes data into groups and aggregates query results based on groups. In this case, GROUP BY returns only one row for each data group. However, window functions, also called online analytical processing (OLAP) functions, can return multiple rows for each data group without aggregating query results. This is different from the traditional GROUP BY function. This topic describes how to use window functions.
Prerequisites
The PolarDB-X 1.0 instance version is 5.4.8 or later.
Limits
Window functions can be used only in SELECT statements.
Window functions cannot be used in conjunction with the separate aggregate functions.
In the following statement, the SUM function that does not include the OVER keyword is an aggregate function. Therefore, this statement cannot be executed.
SELECT SUM(NAME),COUNT() OVER(...) FROM SOME_TABLE
To implement the preceding query, use the following statement:
SELECT SUM(NAME),WIN1 FROM (SELECT NAME,COUNT() OVER(...) AS WIN1 FROM SOME_TABLE) alias
Syntax
function OVER ([[partition by column_some1] [order by column_some2] [RANGE|ROWS BETWEEN start AND end]])
Parameter | Description |
| The window function that you can specify. The following functions are supported:
Note
|
| The partition rule for the window function. This clause divides input rows into different partitions. The process is similar to the division process of the GROUP BY clause. Note You cannot reference complex expressions in the |
| The sorting rule for the window function. This clause defines the order in which the input rows are calculated in the window function. Note You cannot reference complex expressions in the |
| The window frame of the window function. You can use RANGE or ROWS to define the frame. RANGE indicates that the frame is defined by the value range for the computed column. ROWS indicates that the frame is defined by the number of rows for the computed column. You can use the
|
Use cases
Assume that the following raw data has been created.
| year | country | product | profit |
|------|---------|------------|--------|
| 2001 | Finland | Phone | 10 |
| 2000 | Finland | Computer | 1500 |
| 2001 | USA | Calculator | 50 |
| 2001 | USA | Computer | 1500 |
| 2000 | Singapore | Calculator | 75 |
| 2000 | Singapore | Calculator | 75 |
| 2001 | Singapore | Calculator | 79 |
Use the following aggregate function to calculate the total profit of each country:
select country, sum(profit) over (partition by country) sum_profit from test_window;
The following result is returned:
| country | sum_profit | |---------|------------| | Singapore | 229 | | Singapore | 229 | | Singapore | 229 | | USA | 1550 | | USA | 1550 | | Finland | 1510 | | Finland | 1510 |
Use the following dedicated window function to group data by country and rank the products of each country by profit in ascending order:
select 'year', country, product, profit, rank() over (partition by country order by profit) as rank from test_window;
The following result is returned:
| year | country | product | profit | rank | |------|---------|------------|--------|------| | 2001 | Finland | Phone | 10 | 1 | | 2000 | Finland | Computer | 1500 | 2 | | 2001 | USA | Calculator | 50 | 1 | | 2001 | USA | Computer | 1500 | 2 | | 2000 | Singapore | Calculator | 75 | 1 | | 2000 | Singapore | Calculator | 75 | 1 | | 2001 | Singapore | Calculator | 79 | 3 |
Execute the following statement that contains the ROWS option to calculate a cumulative sum of profits for each row in the current window:
select 'year', country, profit, sum(profit) over (partition by country order by 'year' ROWS BETWEEN UNBOUNDED PRECEDING and CURRENT ROW) as sum_win from test_window;
The following result is returned:
+------+---------+--------+-------------+ | year | country | profit | sum_win | +------+---------+--------+-------------+ | 2001 | USA | 50 | 50 | | 2001 | USA | 1500 | 1550 | | 2000 | Singapore | 75 | 75 | | 2000 | Singapore | 75 | 150 | | 2001 | Singapore | 79 | 229 | | 2000 | Finland | 1500 | 1500 | | 2001 | Finland | 10 | 1510 |