Question

这是表格测试的数据：

time       | type 
2017-11-01 | a 
2017-11-01 | a 
2017-11-01 | a 
2017-11-01 | b 
2017-11-02 | a 
2017-11-02 | b 
2017-11-02 | b 
2017-11-02 | b 
2017-11-02 | c 
2017-11-03 | a 
2017-11-03 | a

如何获得结果：

time       | type | count(*)
2017-11-01 | a    | 3
2017-11-01 | b    | 1
2017-11-01 | c    | 0
2017-11-02 | a    | 1
2017-11-02 | b    | 3
2017-11-02 | c    | 1
2017-11-03 | a    | 2
2017-11-03 | b    | 0
2017-11-03 | c    | 0

我尝试使用group by和ifnull，但它无效。我无法从数据中得到0。也许＆＃39;加入＆＃39;可以工作，但如何编辑SQL？

Answer 1

您可以将所有不同类型交叉连接到所有不同的日期（笛卡尔积）

MariaDB [sandbox]> select distinct type,t1.dt from t
    -> cross join (select distinct dt from t) t1;
+------+------------+
| type | dt         |
+------+------------+
| a    | 2017-11-01 |
| b    | 2017-11-01 |
| c    | 2017-11-01 |
| a    | 2017-11-02 |
| b    | 2017-11-02 |
| c    | 2017-11-02 |
| a    | 2017-11-03 |
| b    | 2017-11-03 |
| c    | 2017-11-03 |
+------+------------+
9 rows in set (0.00 sec)

然后离开加入

MariaDB [sandbox]> select s.type,s.dt,ifnull(u.obs,0) obs
    -> from
    -> (
    -> select distinct type,t1.dt from t
    -> cross join (select distinct dt from t) t1
    -> ) s
    -> left join
    -> (select type,dt, count(*) obs from t group by type,dt) u on u.type = s.type and u.dt = s.dt ;
+------+------------+-----+
| type | dt         | obs |
+------+------------+-----+
| a    | 2017-11-01 |   3 |
| b    | 2017-11-01 |   1 |
| c    | 2017-11-01 |   0 |
| a    | 2017-11-02 |   1 |
| b    | 2017-11-02 |   3 |
| c    | 2017-11-02 |   1 |
| a    | 2017-11-03 |   2 |
| b    | 2017-11-03 |   0 |
| c    | 2017-11-03 |   0 |
+------+------------+-----+
9 rows in set (0.00 sec)

请注意，这仅适用于您对某些活动的日期感兴趣的情况 - 如果不考虑构建压缩表并在子查询中使用

mysql：使用group by时如何获得0，结果不存在

1 个答案: