WINDOW_FUNCTION_WINDOW_FUNNEL

WINDOW FUNCTION WINDOW_FUNNEL

description

Searches for event chains in a sliding time window and calculates the maximum number of events that occurred from the chain.

window is the length of time window in seconds.
mode can be one of the followings:
- "default": Defualt mode.
- "deduplication": If the same event holds for the sequence of events, then such repeating event interrupts further processing. E.g. the array parameter is [event1='A', event2='B', event3='C', event4='D'], and the original event chain is "A-B-C-B-D". Since event B repeats, the filtered event chain can only be "A-B-C" and the max event level is 3.
- "fixed": Don't allow interventions of other events. E.g. the array parameter is [event1='A', event2='B', event3='C', event4='D'], and the original event chain is A->B->D->C, it stops finding A->B->C at the D and the max event level is 2.
- "increase": Apply conditions only to events with strictly increasing timestamps.
timestamp_column specifies column of DATETIME type, sliding time window works on it.
evnetN is boolean expression like eventID = 1004.

The function works according to the algorithm:

The function searches for data that triggers the first condition in the chain and sets the event counter to 1. This is the moment when the sliding window starts.
If events from the chain occur sequentially within the window, the counter is incremented. If the sequence of events is disrupted, the counter is not incremented.
If the data has multiple event chains at varying points of completion, the function will only output the size of the longest chain.

window_funnel(window, mode, timestamp_column, event1, event2, ... , eventN)

example

example1: default mode

Using the default mode, find out the maximum number of consecutive events corresponding to different user_id, with a time window of 5 minutes:

CREATE TABLE events(
    user_id BIGINT,
    event_name VARCHAR(64),
    event_timestamp datetime,
    phone_brand varchar(64),
    tab_num int
) distributed by hash(user_id) buckets 3 properties("replication_num" = "1");

INSERT INTO
    events
VALUES
    (100123, 'login', '2022-05-14 10:01:00', 'HONOR', 1),
    (100123, 'visit', '2022-05-14 10:02:00', 'HONOR', 2),
    (100123, 'order', '2022-05-14 10:04:00', 'HONOR', 3),
    (100123, 'payment', '2022-05-14 10:10:00', 'HONOR', 4),
    (100125, 'login', '2022-05-15 11:00:00', 'XIAOMI', 1),
    (100125, 'visit', '2022-05-15 11:01:00', 'XIAOMI', 2),
    (100125, 'order', '2022-05-15 11:02:00', 'XIAOMI', 6),
    (100126, 'login', '2022-05-15 12:00:00', 'IPHONE', 1),
    (100126, 'visit', '2022-05-15 12:01:00', 'HONOR', 2),
    (100127, 'login', '2022-05-15 11:30:00', 'VIVO', 1),
    (100127, 'visit', '2022-05-15 11:31:00', 'VIVO', 5);

SELECT
    user_id,
    window_funnel(
        300,
        "default",
        event_timestamp,
        event_name = 'login',
        event_name = 'visit',
        event_name = 'order',
        event_name = 'payment'
    ) AS level
FROM
    events
GROUP BY
    user_id
order BY
    user_id;

+---------+-------+
| user_id | level |
+---------+-------+
|  100123 |     3 |
|  100125 |     3 |
|  100126 |     2 |
|  100127 |     2 |
+---------+-------+

For uesr_id=100123, because the time when the payment event occurred exceeds the time window, the matched event chain is login-visit-order.

example2: deduplication mode

Use the deduplication mode to find out the maximum number of consecutive events corresponding to different user_ids, with a time window of 1 hour:

CREATE TABLE events(
    user_id BIGINT,
    event_name VARCHAR(64),
    event_timestamp datetime,
    phone_brand varchar(64),
    tab_num int
) distributed by hash(user_id) buckets 3 properties("replication_num" = "1");

INSERT INTO
    events
VALUES
    (100123, 'login', '2022-05-14 10:01:00', 'HONOR', 1),
    (100123, 'visit', '2022-05-14 10:02:00', 'HONOR', 2),
    (100123, 'login', '2022-05-14 10:03:00', 'HONOR', 3),
    (100123, 'order', '2022-05-14 10:04:00', "HONOR", 4),
    (100123, 'payment', '2022-05-14 10:10:00', 'HONOR', 4),
    (100125, 'login', '2022-05-15 11:00:00', 'XIAOMI', 1),
    (100125, 'visit', '2022-05-15 11:01:00', 'XIAOMI', 2),
    (100125, 'order', '2022-05-15 11:02:00', 'XIAOMI', 6),
    (100126, 'login', '2022-05-15 12:00:00', 'IPHONE', 1),
    (100126, 'visit', '2022-05-15 12:01:00', 'HONOR', 2),
    (100127, 'login', '2022-05-15 11:30:00', 'VIVO', 1),
    (100127, 'visit', '2022-05-15 11:31:00', 'VIVO', 5);

SELECT
    user_id,
    window_funnel(
        3600,
        "deduplication",
        event_timestamp,
        event_name = 'login',
        event_name = 'visit',
        event_name = 'order',
        event_name = 'payment'
    ) AS level
FROM
    events
GROUP BY
    user_id
order BY
    user_id;

+---------+-------+
| user_id | level |
+---------+-------+
|  100123 |     2 |
|  100125 |     3 |
|  100126 |     2 |
|  100127 |     2 |
+---------+-------+

For uesr_id=100123, after matching the visit event, the login event appears repeatedly, so the matched event chain is login-visit.

example3: fixed mode

Use the fixed mode to find out the maximum number of consecutive events corresponding to different user_id, with a time window of 1 hour:

CREATE TABLE events(
    user_id BIGINT,
    event_name VARCHAR(64),
    event_timestamp datetime,
    phone_brand varchar(64),
    tab_num int
) distributed by hash(user_id) buckets 3 properties("replication_num" = "1");

INSERT INTO
    events
VALUES
    (100123, 'login', '2022-05-14 10:01:00', 'HONOR', 1),
    (100123, 'visit', '2022-05-14 10:02:00', 'HONOR', 2),
    (100123, 'order', '2022-05-14 10:03:00', "HONOR", 4),
    (100123, 'login2', '2022-05-14 10:04:00', 'HONOR', 3),
    (100123, 'payment', '2022-05-14 10:10:00', 'HONOR', 4),
    (100125, 'login', '2022-05-15 11:00:00', 'XIAOMI', 1),
    (100125, 'visit', '2022-05-15 11:01:00', 'XIAOMI', 2),
    (100125, 'order', '2022-05-15 11:02:00', 'XIAOMI', 6),
    (100126, 'login', '2022-05-15 12:00:00', 'IPHONE', 1),
    (100126, 'visit', '2022-05-15 12:01:00', 'HONOR', 2),
    (100127, 'login', '2022-05-15 11:30:00', 'VIVO', 1),
    (100127, 'visit', '2022-05-15 11:31:00', 'VIVO', 5);

SELECT
    user_id,
    window_funnel(
        3600,
        "fixed",
        event_timestamp,
        event_name = 'login',
        event_name = 'visit',
        event_name = 'order',
        event_name = 'payment'
    ) AS level
FROM
    events
GROUP BY
    user_id
order BY
    user_id;

+---------+-------+
| user_id | level |
+---------+-------+
|  100123 |     3 |
|  100125 |     3 |
|  100126 |     2 |
|  100127 |     2 |
+---------+-------+

For uesr_id=100123, after matching the order event, the event chain is interrupted by the login2 event, so the matched event chain is login-visit-order.

example4: increase mode

Use the increase mode to find out the maximum number of consecutive events corresponding to different user_id, with a time window of 1 hour:

CREATE TABLE events(
    user_id BIGINT,
    event_name VARCHAR(64),
    event_timestamp datetime,
    phone_brand varchar(64),
    tab_num int
) distributed by hash(user_id) buckets 3 properties("replication_num" = "1");

INSERT INTO
    events
VALUES
    (100123, 'login', '2022-05-14 10:01:00', 'HONOR', 1),
    (100123, 'visit', '2022-05-14 10:02:00', 'HONOR', 2),
    (100123, 'order', '2022-05-14 10:04:00', "HONOR", 4),
    (100123, 'payment', '2022-05-14 10:04:00', 'HONOR', 4),
    (100125, 'login', '2022-05-15 11:00:00', 'XIAOMI', 1),
    (100125, 'visit', '2022-05-15 11:01:00', 'XIAOMI', 2),
    (100125, 'order', '2022-05-15 11:02:00', 'XIAOMI', 6),
    (100126, 'login', '2022-05-15 12:00:00', 'IPHONE', 1),
    (100126, 'visit', '2022-05-15 12:01:00', 'HONOR', 2),
    (100127, 'login', '2022-05-15 11:30:00', 'VIVO', 1),
    (100127, 'visit', '2022-05-15 11:31:00', 'VIVO', 5);

SELECT
    user_id,
    window_funnel(
        3600,
        "increase",
        event_timestamp,
        event_name = 'login',
        event_name = 'visit',
        event_name = 'order',
        event_name = 'payment'
    ) AS level
FROM
    events
GROUP BY
    user_id
order BY
    user_id;

+---------+-------+
| user_id | level |
+---------+-------+
|  100123 |     3 |
|  100125 |     3 |
|  100126 |     2 |
|  100127 |     2 |
+---------+-------+

For uesr_id=100123, the timestamp of the payment event and the timestamp of the order event occur in the same second and are not incremented, so the matched event chain is login-visit-order.

keywords

WINDOW,FUNCTION,WINDOW_FUNNEL