在一列上进行 GROUP 并根据另一列中的不同值聚合计数

Question

我有一个像这样的值的表：

CREATE TABLE myTable
    (`ts` varchar(2), `usr_id` varchar(1), `data` varchar(1))
;
    
INSERT INTO myTable
    (`ts`, `usr_id`, `data`)
VALUES
    ('11', 'A', 'x'),
    ('11', 'A', 'x'),
    ('11', 'B', 'x'),
    ('11', 'C', 'x'),
    ('11', 'C', 'x'),
    ('11', 'C', 'x'),
    ('22', 'B', 'x'),
    ('33', 'C', 'x'),
    ('33', 'C', 'x'),
    ('33', 'D', 'x'),
    ('33', 'A', 'x')  
;

我想基于 ts 进行 GROUP BY 并显示 usr_id 的不同值的计数，如下所示。

ts	A的计数	B的计数	其他人的计数
11	2	1	3
22	0	1	0
33	1	0	3

http://www.sqlfiddle.com/#!9/bbf4fc/2

除了为 ts 做 GROUP BY 之外，我没有取得太大进展： select ts, usr_id, count(*) from myTable group by ts, usr_id; 但我想要 A、B 和其他人在同一行的行数，不知道如何实现。

Answer 1

这是您的查询

SELECT 
    DISTINCT ts, 
    COUNT(CASE WHEN usr_id = 'A' THEN 1 END) AS count_a,
    COUNT(CASE WHEN usr_id = 'B' THEN 1 END) AS count_b,
    COUNT(CASE WHEN usr_id != 'A' AND usr_id != 'B' THEN 1 END) AS count_others
FROM myTable
GROUP BY ts
ORDER BY ts

演示

Answer 2

PostgreSQL 允许过滤聚合：

select ts, 
       count(1) filter (where usr_id = 'A') as "count of A",
       count(1) filter (where usr_id = 'B') as "count of B",
       count(1) filter (where usr_id not in ('A', 'B')) as "count of others"
  from mytable
 group by ts;

db<> 在这里摆弄

Answer 3

您的示例链接是 ybout mysql，但是

SELECT
  ts
  , SUM(`usr_id` = 'A') CountA
  , SUM(`usr_id` = 'B') CountB
  , SUM(`usr_id` <> 'B' AND `usr_id` = 'B') CountElse
  
from myTable group by ts

样本

对于 posdgres

select 
  ts
  , SUM(CASE WHEN "usr_id" = 'A' THEN 1 ELSE 0 END) CountA
  , SUM(CASE WHEN "usr_id" = 'B' THEN 1 ELSE 0 END) CountB
  , SUM(CASE WHEN "usr_id" <> 'B' AND "usr_id" = 'B' THEN 1 ELSE 0 END) CountElse
  
  from myTable group by ts

看例子