> ## Documentation Index > Fetch the complete documentation index at: https://docs.risingwave.com/llms.txt > Use this file to discover all available pages before exploring further. # Aggregate functions > Aggregate functions compute a single result from a set of input values. For details about the supported syntaxes of aggregate expressions, see [Aggregate function calls](/sql/query-syntax/value-exp#aggregate-function-calls). ## General-purpose aggregate functions ### `approx_count_distinct` Added in v2.5.0. Counts the estimated number of distinct values in a column, using the HyperLogLog algorithm. This function is only supported for append-only streams, and it is currently in **[technical preview](/changelog/product-lifecycle#product-release-lifecycle)** stage. ```sql theme={null} approx_count_distinct ( expression ) -> bigint ``` ### `array_agg` Collects all the input values, including nulls, into an array. The `ORDER BY` clause is optional and specifies the order of rows processed in the aggregation, which determines the order of the elements in the result array. ```sql theme={null} array_agg ( expression [ ORDER BY sort_expression ] ) -> output_array ``` ### `avg` Returns the average (arithmetic mean) of all non-null input values or null if no non-null values are provided. ```sql theme={null} avg ( expression ) -> see description ``` Input types include smallint, int, bigint, numeric, real, and double precision. Return type is numeric for integer inputs and double precision for float point inputs. ### `bit_and` Returns the bitwise AND of all non-null input values or null if no non-null values are provided. ```sql theme={null} bit_and ( smallint | int | bigint ) -> same as input type ``` ### `bit_or` Returns the bitwise OR of all non-null input values or null if no non-null values are provided. ```sql theme={null} bit_or ( smallint | int | bigint ) -> same as input type ``` ### `bool_and` Returns true if all non-null input values are true, otherwise false. ```sql theme={null} bool_and ( boolean ) -> boolean ``` ### `bool_or` Returns true if any non-null input value is true, otherwise false. ```sql theme={null} bool_or ( boolean ) -> boolean ``` ### `count` Returns the number of non-null input values. ```sql theme={null} count ( expression ) -> bigint ``` The input can be of any supported data type. ### `count(*)` Returns the number of rows in the input. ```sql theme={null} count(*) -> bigint ``` ### `jsonb_agg` Collects all the input values, including nulls, into a JSON array. The `ORDER BY` clause is optional and specifies the order of rows processed in the aggregation, which determines the order of the elements in the result array. ```sql theme={null} jsonb_agg ( any_element [ ORDER BY sort_expression ] ) -> jsonb ``` ### `jsonb_object_agg` Aggregates name/value pairs as a JSON object. Values can be null, but keys cannot. ```sql theme={null} jsonb_object_agg ( key "text" , value "any" ) -> jsonb ``` ### `max` Returns the maximum of the non-null input values, or null if no non-null values are provided. ```sql theme={null} max ( expression ) -> same as input type ``` Input can be of any numeric, string, date/time, or interval type, or an array of these types. ### `min` Returns the minimum value of the non-null input values, or null if no non-null values are provided. ```sql theme={null} min ( expression ) -> same as input type ``` Input can be of any numeric, string, date/time, or interval type, or an array of these types. ### `arg_max` Added in v2.6.0. Returns the first argument where the second argument (comparison value) is the largest. If multiple rows have the same maximum value, use `ORDER BY` to determine which row to return. ```sql theme={null} arg_max (any, any [ORDER BY ...]) -> any ``` ```Example theme={null} select arg_max(name, score) from students; -- Returns the name of the student with the highest score select arg_max(name, score order by age desc) from students; -- Returns the name of the oldest student among those with the highest score ``` ### `arg_min` Added in v2.6.0. Returns the first argument where the second argument (comparison value) is the smallest. If multiple rows have the same minimum value, use `ORDER BY` to determine which row to return. ```sql theme={null} arg_min (any, any [ORDER BY ...]) -> any ``` ```Example theme={null} select arg_min(name, score) from students; -- Returns the name of the student with the lowest score select arg_min(name, score order by age) from students; -- Returns the name of the youngest student among those with the lowest score ``` ### `string_agg` Concatenates non-null input values into a string. Each value after the first is preceded by the corresponding delimiter (if it's not null). If no non-null values are provided, returns null. The `ORDER BY` clause is optional and specifies the order of rows processed in the aggregation, which determines the order of the elements in the result array. ```sql theme={null} string_agg ( value text, delimiter text [ ORDER BY sort_expression ] ) -> output_string ``` ### `sum` Returns the sum of all non-null input values, or null if no non-null values are provided. ```sql theme={null} sum ( expression ) ``` Input types include smallint, int, bigint, numeric, real, and double precision. Return type is bigint for smallint or int inputs, numeric for bigint inputs, otherwise the same as the input data type. ### `first_value` Returns the first value in an ordered set of values, including nulls. ```sql theme={null} first_value ( expression ORDER BY order_key ) -> same as input type ``` `order_key` is the column or expression used to determine the order of the values. It is required to make the result deterministic. ### `last_value` Returns the last value in an ordered set of values, including nulls. ```sql theme={null} last_value ( expression ORDER BY order_key ) -> same as input type ``` ## Aggregate functions for statistics ### `stddev_pop` Calculates the population standard deviation of the input values. Returns `NULL` if the input contains no non-null values. ```sql theme={null} stddev_pop ( expression ) -> output_value ``` ### `stddev_samp` Calculates the sample standard deviation of the input values. Returns `NULL` if the input contains fewer than two non-null values. ```sql theme={null} stddev_samp ( expression ) -> output_value ``` ### `var_pop` Calculates the population variance of the input values. Returns `NULL` if the input contains no non-null values. ```sql theme={null} var_pop ( expression ) -> output_value ``` ### `var_samp` Calculates the sample variance of the input values. Returns `NULL` if the input contains fewer than two non-null values. ```sql theme={null} var_samp ( expression ) -> output_value ``` ## Ordered-set aggregate functions At present, ordered-set aggregate functions support only constant fraction arguments. ### `mode` Computes the mode, which is the most frequent value of the aggregated argument. If there are multiple equally-frequent values, it arbitrarily chooses the first one. ```sql theme={null} mode () WITHIN GROUP ( ORDER BY sort_expression ) -> same as sort_expression ``` `sort_expression`: Must be of a sortable type. This example calculates the mode of the values in `column1` from `table1`. ```sql theme={null} SELECT mode() WITHIN GROUP (ORDER BY column1) FROM table1; ``` ### `percentile_cont` At present, `percentile_cont` is not supported for [streaming queries](/reference/key-concepts#streaming-queries) yet. Computes the continuous percentile, which is a value corresponding to the specified fraction within the ordered set of aggregated argument values. It can interpolate between adjacent input items if needed. ```sql theme={null} percentile_cont ( fraction double precision ) WITHIN GROUP ( ORDER BY sort_expression double precision ) -> double precision ``` `fraction`: The fraction value representing the desired percentile. It should be between 0 and 1. This example calculates the median (50th percentile) of the values in `column1` from `table1`. ```sql theme={null} SELECT percentile_cont(0.5) WITHIN GROUP (ORDER BY column1) FROM table1; ``` If NULL is provided, the function will not calculate a specific percentile and return NULL instead. ### `percentile_disc` At present, `percentile_disc` is not supported for streaming queries yet. Computes the discrete percentile, which is the first value within the ordered set of aggregated argument values whose position in the ordering equals or exceeds the specified fraction. ```sql theme={null} percentile_disc ( fraction double precision ) WITHIN GROUP ( ORDER BY sort_expression anyelement ) -> same as sort_expression ``` `fraction`: The fraction value representing the desired percentile. It should be between 0 and 1. `sort_expression`: Must be of a sortable type. This example calculates the 75th percentile of the values in `column1` from `table1`. ```sql theme={null} SELECT percentile_disc(0.75) WITHIN GROUP (ORDER BY column1) FROM table1; ``` If NULL is provided, the function will not calculate a specific percentile and return NULL instead. ### `approx_percentile` Returns an approximate value of the specified percentile from a numeric column. ```sql theme={null} approx_percentile(DOUBLE percentile [, DOUBLE relative_error]) WITHIN GROUP (ORDER BY percentile_column) -> output_value ``` * `percentile`: The percentile to approximate. It should be between 0 and 1. * `relative_error`: Optional. Specifies the maximum allowed error in the approximation. Defaults to 0.01 (1%). * `percentile_column`: The column from which to calculate the percentile. Must be of a numeric type. This example calculates the 50th percentile of a numeric column with the default relative error: ```sql theme={null} SELECT approx_percentile(0.5) WITHIN GROUP (ORDER BY column1) FROM table1; ``` ## Grouping operation functions Grouping operation functions are used in conjunction with grouping sets to distinguish result rows. The arguments to the `grouping()` function are not actually evaluated, but they must exactly match expressions given in the `GROUP BY` clause of the associated query level. ### `grouping` Returns a bit mask indicating which `GROUP BY` expressions are not included in the current grouping set. Bits are assigned with the rightmost argument corresponding to the least-significant bit; each bit is 0 if the corresponding expression is included in the grouping criteria of the grouping set generating the current result row, and 1 if it is not included. ```sql Syntax theme={null} grouping ( group_by_expression(s) ) → integer ``` #### Example ```sql Create a table theme={null} CREATE TABLE items_sold (brand varchar, size varchar, sales int); ``` ```sql Insert some data theme={null} INSERT INTO items_sold VALUES ('Foo', 'L', 10),('Foo', 'M', 20),('Bar', 'M', 15),('Bar', 'L', '5'); ``` ```sql Get grouping results theme={null} SELECT brand, size, sum(sales), grouping(brand), grouping(size), grouping(brand,size), count(DISTINCT sales) FROM items_sold GROUP BY GROUPING SETS ((brand), (size), ()); ------RESULTS Bar NULL 20 0 1 1 2 Foo NULL 30 0 1 1 2 NULL L 15 1 0 2 2 NULL M 35 1 0 2 2 NULL NULL 50 1 1 3 4 ```