Compartilhamento de tecnologia

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Dados de teste

-- 创建 user_activity 表
DROP TABLE IF EXISTS user_activity ;
CREATE TABLE user_activity (
    user_id STRING,
    activity_start TIMESTAMP,
    activity_end TIMESTAMP
);

-- 插入数据
INSERT INTO user_activity VALUES
('user1', '2024-07-11 08:00:00', '2024-07-11 09:00:00'),
('user2', '2024-07-11 08:30:00', '2024-07-11 09:30:00'),
('user3', '2024-07-11 09:00:00', '2024-07-11 10:00:00'),
('user4', '2024-07-11 09:15:00', '2024-07-11 09:45:00'),
('user5', '2024-07-11 09:30:00', '2024-07-11 10:30:00'),
('user6', '2024-07-11 10:00:00', '2024-07-11 11:00:00'),
('user7', '2024-07-11 08:05:00', '2024-07-11 08:55:00'),
('user8', '2024-07-11 08:45:00', '2024-07-11 09:15:00'),
('user9', '2024-07-11 09:05:00', '2024-07-11 10:05:00'),
('user10', '2024-07-11 09:25:00', '2024-07-11 10:25:00'),
('user11', '2024-07-11 08:10:00', '2024-07-11 09:10:00'),
('user12', '2024-07-11 08:20:00', '2024-07-11 09:20:00'),
('user13', '2024-07-11 08:35:00', '2024-07-11 09:35:00'),
('user14', '2024-07-11 08:50:00', '2024-07-11 09:50:00'),
('user15', '2024-07-11 09:10:00', '2024-07-11 10:10:00'),
('user16', '2024-07-11 09:20:00', '2024-07-11 10:20:00'),
('user17', '2024-07-11 09:40:00', '2024-07-11 10:40:00'),
('user18', '2024-07-11 10:05:00', '2024-07-11 11:05:00'),
('user19', '2024-07-11 10:15:00', '2024-07-11 11:15:00'),
('user20', '2024-07-11 10:25:00', '2024-07-11 11:25:00');
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

Declaração de necessidades

Calcule o número máximo de pessoas online em cada momento para um determinado sistema.

Exemplo de resultados:

Imprensa de resultado activity_time Ascendente.

em:

• activity_time Indica o ponto estatístico no tempo;
• max_users Indica o maior número de pico de pessoas naquele momento.

Realização de requisitos

select
    date_format(activity_time,'yyyy-MM-dd HH') activity_time,
    max(total_users) max_users
from
    (select
        activity_time,
        sum(flag) over(order by activity_time) total_users
    from
        (select
            activity_start activity_time,
            1 flag
        from
            user_activity
        union all
        select
            activity_end activity_time,
            -1 flag
        from
            user_activity)t1
    )t2
group by
    date_format(activity_time,'yyyy-MM-dd HH');
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

A saída é a seguinte:

O núcleo desta questão é a subconsulta t2 A lógica em:

	select
        activity_time,
        sum(flag) over(order by activity_time) total_users
    from
        (select
            activity_start activity_time,
            1 flag
        from
            user_activity
        union all
        select
            activity_end activity_time,
            -1 flag
        from
            user_activity)t1;
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

Primeiro, subconsultamos t1 Converta colunas em linhas, então por que você precisa fazer isso? Claro que isto é por conveniência estatística.

Vamos pensar sobre isso. Quando um usuário faz login e entra no sistema, o número de pessoas no sistema aumentará? +1, pelo contrário, quando o usuário sair, o número de pessoas aumentará? -1。

Quando colocamos os tempos de login e logout na mesma coluna e classificamos por tempo, podemos calcular com precisão o número de pessoas online em cada momento? t2 O que fazemos é realizar cálculos cumulativos através de funções de janela,t2 O resultado é assim:

2024-07-11 08:00:00     1
2024-07-11 08:05:00     2
2024-07-11 08:10:00     3
2024-07-11 08:20:00     4
2024-07-11 08:30:00     5
2024-07-11 08:35:00     6
2024-07-11 08:45:00     7
2024-07-11 08:50:00     8
2024-07-11 08:55:00     7
2024-07-11 09:00:00     7
2024-07-11 09:00:00     7
2024-07-11 09:05:00     8
2024-07-11 09:10:00     8
2024-07-11 09:10:00     8
2024-07-11 09:15:00     8
2024-07-11 09:15:00     8
2024-07-11 09:20:00     8
2024-07-11 09:20:00     8
2024-07-11 09:25:00     9
2024-07-11 09:30:00     9
2024-07-11 09:30:00     9
2024-07-11 09:35:00     8
2024-07-11 09:40:00     9
2024-07-11 09:45:00     8
2024-07-11 09:50:00     7
2024-07-11 10:00:00     7
2024-07-11 10:00:00     7
2024-07-11 10:05:00     7
2024-07-11 10:05:00     7
2024-07-11 10:10:00     6
2024-07-11 10:15:00     7
2024-07-11 10:20:00     6
2024-07-11 10:25:00     6
2024-07-11 10:25:00     6
2024-07-11 10:30:00     5
2024-07-11 10:40:00     4
2024-07-11 11:00:00     3
2024-07-11 11:05:00     2
2024-07-11 11:15:00     1
2024-07-11 11:25:00     0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40

Por fim, agrupe e agregue por pontos no tempo, por meio de max A função encontra o número máximo de pico de pessoas em cada momento e conclui ~