Categories:

Window Functions (General)

CONDITIONAL_CHANGE_EVENT

Returns a window event number for each row within a window partition when the value of the argument expr1 in the current row is different from the value of expr1 in the previous row. The window event number starts from 0 and is incremented by 1 to indicate the number of changes so far within that window.

Syntax

CONDITIONAL_CHANGE_EVENT( <expr1> ) OVER ( [ PARTITION BY <expr2> ] ORDER BY <expr3> )

Arguments

expr1

This is an expression that gets compared with the expression of the previous row.

expr2

This is the optional expression to partition by.

expr3

This is the expression to order by within each partition.

Usage Notes

  • The expression CONDITIONAL_CHANGE_EVENT (expr1) OVER (window_frame) is calculated as:

    CONDITIONAL_TRUE_EVENT( <expr1> != LAG(<expr1>) OVER(window_frame)) OVER(window_frame)

    For more information about CONDITIONAL_TRUE_EVENT, see CONDITIONAL_TRUE_EVENT.

Examples

This shows how to detect the number of times that the power failed and was turned back on (i.e. the number of times that the voltage dropped to 0 or was restored). (This example assumes that sampling the voltage every 15 minutes is sufficient. Since power failures can last less than 15 minutes, you’d typically want more frequent samples, or you’d want to treat the query results as an approximation.)

Create and load the table:

CREATE TABLE voltage_readings (
    site_ID INTEGER, -- which refrigerator the measurement was taken in.
    ts TIMESTAMP,  -- the time at which the temperature was measured.
    VOLTAGE FLOAT
    );
INSERT INTO voltage_readings (site_ID, ts, voltage) VALUES
    (1, '2019-10-30 13:00:00', 120),
    (1, '2019-10-30 13:15:00', 120),
    (1, '2019-10-30 13:30:00',   0),
    (1, '2019-10-30 13:45:00',   0),
    (1, '2019-10-30 14:00:00',   0),
    (1, '2019-10-30 14:15:00',   0),
    (1, '2019-10-30 14:30:00', 120)
    ;

This shows the samples for which the voltage was zero, whether or not those zero-volt events were part of the same power failure or different power failures.

SELECT site_ID, ts, voltage
    FROM voltage_readings
    WHERE voltage = 0;
+---------+-------------------------+---------+
| SITE_ID | TS                      | VOLTAGE |
|---------+-------------------------+---------|
|       1 | 2019-10-30 13:30:00.000 |       0 |
|       1 | 2019-10-30 13:45:00.000 |       0 |
|       1 | 2019-10-30 14:00:00.000 |       0 |
|       1 | 2019-10-30 14:15:00.000 |       0 |
+---------+-------------------------+---------+

This shows the samples, along with a column indicating whether the voltage changed:

SELECT
      site_ID,
      ts,
      voltage,
      CONDITIONAL_CHANGE_EVENT(voltage = 0) OVER (ORDER BY ts) AS power_changes
    FROM voltage_readings;
+---------+-------------------------+---------+---------------+
| SITE_ID | TS                      | VOLTAGE | POWER_CHANGES |
|---------+-------------------------+---------+---------------|
|       1 | 2019-10-30 13:00:00.000 |     120 |             0 |
|       1 | 2019-10-30 13:15:00.000 |     120 |             0 |
|       1 | 2019-10-30 13:30:00.000 |       0 |             1 |
|       1 | 2019-10-30 13:45:00.000 |       0 |             1 |
|       1 | 2019-10-30 14:00:00.000 |       0 |             1 |
|       1 | 2019-10-30 14:15:00.000 |       0 |             1 |
|       1 | 2019-10-30 14:30:00.000 |     120 |             2 |
+---------+-------------------------+---------+---------------+

This shows the times that the power stopped and restarted:

WITH power_change_events AS
    (
    SELECT
      site_ID,
      ts,
      voltage,
      CONDITIONAL_CHANGE_EVENT(voltage = 0) OVER (ORDER BY ts) AS power_changes
    FROM voltage_readings
    )
SELECT
      site_ID,
      MIN(ts),
      voltage,
      power_changes
    FROM power_change_events
    GROUP BY site_ID, power_changes, voltage
    ;
+---------+-------------------------+---------+---------------+
| SITE_ID | MIN(TS)                 | VOLTAGE | POWER_CHANGES |
|---------+-------------------------+---------+---------------|
|       1 | 2019-10-30 13:00:00.000 |     120 |             0 |
|       1 | 2019-10-30 13:30:00.000 |       0 |             1 |
|       1 | 2019-10-30 14:30:00.000 |     120 |             2 |
+---------+-------------------------+---------+---------------+

This shows how many times the power stopped and restarted:

WITH power_change_events AS
    (
    SELECT
          site_ID,
          CONDITIONAL_CHANGE_EVENT(voltage = 0) OVER (ORDER BY ts) AS power_changes
        FROM voltage_readings
    )
SELECT MAX(power_changes) 
    FROM power_change_events
    GROUP BY site_ID
    ;
+--------------------+
| MAX(POWER_CHANGES) |
|--------------------|
|                  2 |
+--------------------+

This example illustrates that:

  • The change number within a partition changes each time the specified value changes.

  • NULL values are not considered a new or changed value.

  • The change count starts over at 0 for each partition.

    Create and load the table:

    CREATE TABLE table1 (province VARCHAR, o_col INTEGER, o2_col INTEGER);
    INSERT INTO table1 (province, o_col, o2_col) VALUES
        ('Alberta',    0, 10),
        ('Alberta',    0, 10),
        ('Alberta',   13, 10),
        ('Alberta',   13, 11),
        ('Alberta',   14, 11),
        ('Alberta',   15, 12),
        ('Alberta', NULL, NULL),
        ('Manitoba',    30, 30);
    

    Query the table:

    SELECT province, o_col,
          CONDITIONAL_CHANGE_EVENT(o_col) 
            OVER (PARTITION BY province ORDER BY o_col) 
              AS change_event
        FROM table1
        ORDER BY province, o_col
        ;
    +----------+-------+--------------+
    | PROVINCE | O_COL | CHANGE_EVENT |
    |----------+-------+--------------|
    | Alberta  |     0 |            0 |
    | Alberta  |     0 |            0 |
    | Alberta  |    13 |            1 |
    | Alberta  |    13 |            1 |
    | Alberta  |    14 |            2 |
    | Alberta  |    15 |            3 |
    | Alberta  |  NULL |            3 |
    | Manitoba |    30 |            0 |
    +----------+-------+--------------+
    

The next example shows that:

  • expr1 can be an expression other than a column. This query uses the expression o_col < 15, and the output of the query shows when the value in o_col changes from a value less than 15 to a value greater than or equal to 15.

  • expr3 does not need to match expr1. In other words, the expression in the ORDER BY sub-clause of the OVER clause does not need to match the expression in the CONDITIONAL_CHANGE_EVENT function.

    Query the table:

    SELECT province, o_col,
          'o_col < 15' AS condition,
          CONDITIONAL_CHANGE_EVENT(o_col) 
            OVER (PARTITION BY province ORDER BY o_col) 
              AS change_event,
          CONDITIONAL_CHANGE_EVENT(o_col < 15) 
            OVER (PARTITION BY province ORDER BY o_col) 
              AS change_event_2
        FROM table1
        ORDER BY province, o_col
        ;
    +----------+-------+------------+--------------+----------------+
    | PROVINCE | O_COL | CONDITION  | CHANGE_EVENT | CHANGE_EVENT_2 |
    |----------+-------+------------+--------------+----------------|
    | Alberta  |     0 | o_col < 15 |            0 |              0 |
    | Alberta  |     0 | o_col < 15 |            0 |              0 |
    | Alberta  |    13 | o_col < 15 |            1 |              0 |
    | Alberta  |    13 | o_col < 15 |            1 |              0 |
    | Alberta  |    14 | o_col < 15 |            2 |              0 |
    | Alberta  |    15 | o_col < 15 |            3 |              1 |
    | Alberta  |  NULL | o_col < 15 |            3 |              1 |
    | Manitoba |    30 | o_col < 15 |            0 |              0 |
    +----------+-------+------------+--------------+----------------+
    

The next example compares CONDITIONAL_CHANGE_EVENT and CONDITIONAL_TRUE_EVENT:

SELECT province, o_col,
      CONDITIONAL_CHANGE_EVENT(o_col) 
        OVER (PARTITION BY province ORDER BY o_col) 
          AS change_event,
      CONDITIONAL_TRUE_EVENT(o_col) 
        OVER (PARTITION BY province ORDER BY o_col) 
          AS true_event
    FROM table1
    ORDER BY province, o_col
    ;
+----------+-------+--------------+------------+
| PROVINCE | O_COL | CHANGE_EVENT | TRUE_EVENT |
|----------+-------+--------------+------------|
| Alberta  |     0 |            0 |          0 |
| Alberta  |     0 |            0 |          0 |
| Alberta  |    13 |            1 |          1 |
| Alberta  |    13 |            1 |          2 |
| Alberta  |    14 |            2 |          3 |
| Alberta  |    15 |            3 |          4 |
| Alberta  |  NULL |            3 |          4 |
| Manitoba |    30 |            0 |          1 |
+----------+-------+--------------+------------+

This example also compares CONDITIONAL_CHANGE_EVENT and CONDITIONAL_TRUE_EVENT:

CREATE TABLE borrowers (
    name VARCHAR,
    status_date DATE,
    late_balance NUMERIC(11, 2),
    thirty_day_late_balance NUMERIC(11, 2)
    );
INSERT INTO borrowers (name, status_date, late_balance, thirty_day_late_balance) VALUES

    -- Pays late frequently, but catches back up rather than falling further
    -- behind.
    ('Geoffrey Flake', '2018-01-01'::DATE,    0.0,    0.0),
    ('Geoffrey Flake', '2018-02-01'::DATE, 1000.0,    0.0),
    ('Geoffrey Flake', '2018-03-01'::DATE, 2000.0, 1000.0),
    ('Geoffrey Flake', '2018-04-01'::DATE,    0.0,    0.0),
    ('Geoffrey Flake', '2018-05-01'::DATE, 1000.0,    0.0),
    ('Geoffrey Flake', '2018-06-01'::DATE, 2000.0, 1000.0),
    ('Geoffrey Flake', '2018-07-01'::DATE,    0.0,    0.0),
    ('Geoffrey Flake', '2018-08-01'::DATE,    0.0,    0.0),

    -- Keeps falling further behind.
    ('Cy Dismal', '2018-01-01'::DATE,    0.0,    0.0),
    ('Cy Dismal', '2018-02-01'::DATE,    0.0,    0.0),
    ('Cy Dismal', '2018-03-01'::DATE, 1000.0,    0.0),
    ('Cy Dismal', '2018-04-01'::DATE, 2000.0, 1000.0),
    ('Cy Dismal', '2018-05-01'::DATE, 3000.0, 2000.0),
    ('Cy Dismal', '2018-06-01'::DATE, 4000.0, 3000.0),
    ('Cy Dismal', '2018-07-01'::DATE, 5000.0, 4000.0),
    ('Cy Dismal', '2018-08-01'::DATE, 6000.0, 5000.0),

    -- Fell behind and isn't catching up, but isn't falling further and 
    -- further behind. Essentially, this person just 'failed' once.
    ('Leslie Safer', '2018-01-01'::DATE,    0.0,    0.0),
    ('Leslie Safer', '2018-02-01'::DATE,    0.0,    0.0),
    ('Leslie Safer', '2018-03-01'::DATE, 1000.0, 1000.0),
    ('Leslie Safer', '2018-04-01'::DATE, 2000.0, 1000.0),
    ('Leslie Safer', '2018-05-01'::DATE, 2000.0, 1000.0),
    ('Leslie Safer', '2018-06-01'::DATE, 2000.0, 1000.0),
    ('Leslie Safer', '2018-07-01'::DATE, 2000.0, 1000.0),
    ('Leslie Safer', '2018-08-01'::DATE, 2000.0, 1000.0),

    -- Always pays on time and in full.
    ('Ida Idyll', '2018-01-01'::DATE,    0.0,    0.0),
    ('Ida Idyll', '2018-02-01'::DATE,    0.0,    0.0),
    ('Ida Idyll', '2018-03-01'::DATE,    0.0,    0.0),
    ('Ida Idyll', '2018-04-01'::DATE,    0.0,    0.0),
    ('Ida Idyll', '2018-05-01'::DATE,    0.0,    0.0),
    ('Ida Idyll', '2018-06-01'::DATE,    0.0,    0.0),
    ('Ida Idyll', '2018-07-01'::DATE,    0.0,    0.0),
    ('Ida Idyll', '2018-08-01'::DATE,    0.0,    0.0)

    ;
SELECT name, status_date, late_balance AS "OVERDUE", 
        thirty_day_late_balance AS "30 DAYS OVERDUE",
        CONDITIONAL_CHANGE_EVENT(thirty_day_late_balance) 
          OVER (PARTITION BY name ORDER BY status_date) AS change_event_cnt,
        CONDITIONAL_TRUE_EVENT(thirty_day_late_balance) 
          OVER (PARTITION BY name ORDER BY status_date) AS true_cnt
    FROM borrowers
    ORDER BY name, status_date
    ;
+----------------+-------------+---------+-----------------+------------------+----------+
| NAME           | STATUS_DATE | OVERDUE | 30 DAYS OVERDUE | CHANGE_EVENT_CNT | TRUE_CNT |
|----------------+-------------+---------+-----------------+------------------+----------|
| Cy Dismal      | 2018-01-01  |    0.00 |            0.00 |                0 |        0 |
| Cy Dismal      | 2018-02-01  |    0.00 |            0.00 |                0 |        0 |
| Cy Dismal      | 2018-03-01  | 1000.00 |            0.00 |                0 |        0 |
| Cy Dismal      | 2018-04-01  | 2000.00 |         1000.00 |                1 |        1 |
| Cy Dismal      | 2018-05-01  | 3000.00 |         2000.00 |                2 |        2 |
| Cy Dismal      | 2018-06-01  | 4000.00 |         3000.00 |                3 |        3 |
| Cy Dismal      | 2018-07-01  | 5000.00 |         4000.00 |                4 |        4 |
| Cy Dismal      | 2018-08-01  | 6000.00 |         5000.00 |                5 |        5 |
| Geoffrey Flake | 2018-01-01  |    0.00 |            0.00 |                0 |        0 |
| Geoffrey Flake | 2018-02-01  | 1000.00 |            0.00 |                0 |        0 |
| Geoffrey Flake | 2018-03-01  | 2000.00 |         1000.00 |                1 |        1 |
| Geoffrey Flake | 2018-04-01  |    0.00 |            0.00 |                2 |        1 |
| Geoffrey Flake | 2018-05-01  | 1000.00 |            0.00 |                2 |        1 |
| Geoffrey Flake | 2018-06-01  | 2000.00 |         1000.00 |                3 |        2 |
| Geoffrey Flake | 2018-07-01  |    0.00 |            0.00 |                4 |        2 |
| Geoffrey Flake | 2018-08-01  |    0.00 |            0.00 |                4 |        2 |
| Ida Idyll      | 2018-01-01  |    0.00 |            0.00 |                0 |        0 |
| Ida Idyll      | 2018-02-01  |    0.00 |            0.00 |                0 |        0 |
| Ida Idyll      | 2018-03-01  |    0.00 |            0.00 |                0 |        0 |
| Ida Idyll      | 2018-04-01  |    0.00 |            0.00 |                0 |        0 |
| Ida Idyll      | 2018-05-01  |    0.00 |            0.00 |                0 |        0 |
| Ida Idyll      | 2018-06-01  |    0.00 |            0.00 |                0 |        0 |
| Ida Idyll      | 2018-07-01  |    0.00 |            0.00 |                0 |        0 |
| Ida Idyll      | 2018-08-01  |    0.00 |            0.00 |                0 |        0 |
| Leslie Safer   | 2018-01-01  |    0.00 |            0.00 |                0 |        0 |
| Leslie Safer   | 2018-02-01  |    0.00 |            0.00 |                0 |        0 |
| Leslie Safer   | 2018-03-01  | 1000.00 |         1000.00 |                1 |        1 |
| Leslie Safer   | 2018-04-01  | 2000.00 |         1000.00 |                1 |        2 |
| Leslie Safer   | 2018-05-01  | 2000.00 |         1000.00 |                1 |        3 |
| Leslie Safer   | 2018-06-01  | 2000.00 |         1000.00 |                1 |        4 |
| Leslie Safer   | 2018-07-01  | 2000.00 |         1000.00 |                1 |        5 |
| Leslie Safer   | 2018-08-01  | 2000.00 |         1000.00 |                1 |        6 |
+----------------+-------------+---------+-----------------+------------------+----------+

Here is a more extensive example:

CREATE OR REPLACE TABLE table tbl
(p int, o int, i int, r int, s varchar(100));

INSERT INTO tbl VALUES
(100,1,1,70,'seventy'),(100,2,2,30, 'thirty'),(100,3,3,40,'fourty'),(100,4,NULL,90,'ninety'),(100,5,5,50,'fifty'),(100,6,6,30,'thirty'),
(200,7,7,40,'fourty'),(200,8,NULL,NULL,'n_u_l_l'),(200,9,NULL,NULL,'n_u_l_l'),(200,10,10,20,'twenty'),(200,11,NULL,90,'ninety'),
(300,12,12,30,'thirty'),
(400,13,NULL,20,'twenty');

SELECT * FROM tbl ORDER BY p, o, i;

+-----+----+--------+--------+---------+
|  P  | O  |   I    |   R    |    S    |
+-----+----+--------+--------+---------+
| 100 | 1  | 1      | 70     | seventy |
| 100 | 2  | 2      | 30     | thirty  |
| 100 | 3  | 3      | 40     | fourty  |
| 100 | 4  | [NULL] | 90     | ninety  |
| 100 | 5  | 5      | 50     | fifty   |
| 100 | 6  | 6      | 30     | thirty  |
| 200 | 7  | 7      | 40     | fourty  |
| 200 | 8  | [NULL] | [NULL] | n_u_l_l |
| 200 | 9  | [NULL] | [NULL] | n_u_l_l |
| 200 | 10 | 10     | 20     | twenty  |
| 200 | 11 | [NULL] | 90     | ninety  |
| 300 | 12 | 12     | 30     | thirty  |
| 400 | 13 | [NULL] | 20     | twenty  |
+-----+----+--------+--------+---------+

SELECT p, o, CONDITIONAL_CHANGE_EVENT(o) OVER (PARTITION BY p ORDER BY o) FROM tbl ORDER BY p, o;

+-----+----+--------------------------------------------------------------+
|   P |  O | CONDITIONAL_CHANGE_EVENT(O) OVER (PARTITION BY P ORDER BY O) |
|-----+----+--------------------------------------------------------------|
| 100 |  1 |                                                            0 |
| 100 |  2 |                                                            1 |
| 100 |  3 |                                                            2 |
| 100 |  4 |                                                            3 |
| 100 |  5 |                                                            4 |
| 100 |  6 |                                                            5 |
| 200 |  7 |                                                            0 |
| 200 |  8 |                                                            1 |
| 200 |  9 |                                                            2 |
| 200 | 10 |                                                            3 |
| 200 | 11 |                                                            4 |
| 300 | 12 |                                                            0 |
| 400 | 13 |                                                            0 |
+-----+----+--------------------------------------------------------------+