MySQL-最大计数多组

Question

我有如下表：

╒═══════════╤════════════╤════════════╕ 
│ iddmp2g_2 │ A_BCF_name │ A_BTS_name │ 
╞═══════════╪════════════╪════════════╡ 
│ 1   │ CAR.Inem │ CAR.Inem_A │ 
├───────────┼────────────┼────────────┤ 
│ 2   │ CAR.Inem │ CAR.Inem_A │ 
├───────────┼────────────┼────────────┤ 
│ 3   │ CAR.Inem │ CAR.Inem_A │ 
├───────────┼────────────┼────────────┤ 
│ 4   │ CAR.Inem │ CAR.Inem_A │ 
├───────────┼────────────┼────────────┤ 
│ 5   │ CAR.Inem │ CAR.Inem_B │ 
├───────────┼────────────┼────────────┤ 
│ 6   │ CAR.Inem │ CAR.Inem_B │ 
├───────────┼────────────┼────────────┤ 
│ 7   │ CAR.Inem │ CAR.Inem_B │ 
├───────────┼────────────┼────────────┤ 
│ 8   │ CAR.Inem │ CAR.Inem_1 │ 
├───────────┼────────────┼────────────┤ 
│ 9   │ CAR.Inem │ CAR.Inem_1 │ 
├───────────┼────────────┼────────────┤ 
│ 10  │ CAR.Inem │ CAR.Inem_1 │ 
├───────────┼────────────┼────────────┤ 
│ 11  │ CAR.Inem │ CAR.Inem_2 │ 
├───────────┼────────────┼────────────┤ 
│ 12  │ CAR.Inem │ CAR.Inem_2 │ 
└───────────┴────────────┴────────────┘ 

如果我申请以下查询表中我得到：

SELECT A_BCF_name, COUNT(A_BTS_name) from dmp2g_2 
group by A_BCF_name, A_BTS_name 

结果：

╔════════════╤════════════╤═══════════════════╗ 
║ A_BCF_name │ A_BTS_name │ COUNT(A_BTS_name) ║ 
╠════════════╪════════════╪═══════════════════╣ 
║ CAR.Inem │ CAR.Inem_A │ 4     ║ 
╟────────────┼────────────┼───────────────────╢ 
║ CAR.Inem │ CAR.Inem_B │ 3     ║ 
╟────────────┼────────────┼───────────────────╢ 
║ CAR.Inem │ CAR.Inem_1 │ 3     ║ 
╟────────────┼────────────┼───────────────────╢ 
║ CAR.Inem │ CAR.Inem_2 │ 2     ║ 
╚════════════╧════════════╧═══════════════════╝ 

接下来的方法是根据最后一个字符（数字或字母）得到两组A_BTS_name，如下所示：

SELECT A_BCF_name, A_BTS_name, COUNT(A_BTS_name) 
FROM dmp2g_2 
WHERE SUBSTRING_INDEX(A_BTS_name,'_',-1) REGEXP '[0-9]+' /*getting only records that end in a number*/ 
GROUP BY A_BCF_name, A_BTS_name 

输出：在一个字母或字符结束

╔════════════╤════════════╤═══════════════════╗ 
║ A_BCF_name │ A_BTS_name │ COUNT(A_BTS_name) ║ 
╠════════════╪════════════╪═══════════════════╣ 
║ CAR.Inem │ CAR.Inem_1 │ 3     ║ 
╟────────────┼────────────┼───────────────────╢ 
║ CAR.Inem │ CAR.Inem_2 │ 2     ║ 
╚════════════╧════════════╧═══════════════════╝ 

获取记录。

SELECT A_BCF_name, A_BTS_name, COUNT(A_BTS_name) 
FROM dmp2g_2 
WHERE SUBSTRING_INDEX(A_BTS_name,'_',-1) REGEXP '[A-Za-z]' /*filter records that ends in a letter*/ 
GROUP BY A_BCF_name, A_BTS_name 

输出：

╔════════════╤════════════╤═══════════════════╗ 
║ A_BCF_name │ A_BTS_name │ COUNT(A_BTS_name) ║ 
╠════════════╪════════════╪═══════════════════╣ 
║ CAR.Inem │ CAR.Inem_A │ 4     ║ 
╟────────────┼────────────┼───────────────────╢ 
║ CAR.Inem │ CAR.Inem_B │ 3     ║ 
+────────────┼────────────┼───────────────────+ 

此时一切正常。

我的问题是如何获得每种组的最大数量？我需要在CAR.Inem_A和CAR.Inem_B之间获得最大数量，并在CAR.Inem_1和CAR.Inem_2之间获得最大数量。结果应该是这样的：

╔════════════╤════════════════════════╤════════════════════════╗ 
║ A_BCF_name │ MAX(A_BTS_name_number) │ MAX(A_BTS_name_letter) ║ 
╠════════════╪════════════════════════╪════════════════════════╣ 
║ CAR.Inem │ 3      │ 4      ║ 
╚════════════╧════════════════════════╧════════════════════════╝ 

我感谢所有帮助

感谢

Answer 1

这里是你问的，让你的信息的查询，但结果被安排在一个更查询 - 友好的方式。

select 
t3.A_BCF_name, 
t3.record_type, 
max(t3.record_count) as max_record_count 
from 
    (
    select 
    t2.A_BCF_name, 
    t2.A_BTS_name, 
    t2.record_type, 
    count(t2.record_type) record_count 
    from 
    (
    select 
    t1.iddmp2g_2, 
    t1.A_BCF_name, 
    t1.A_BTS_name, 
    case when SUBSTRING_INDEX(t1.A_BTS_name,'_',-1) REGEXP '[A-Za-z]' then 'letter' else 'number' end as record_type 
    from dmp2g_2 as t1 
    ) as t2 
    group by t2.A_BCF_name, t2.A_BTS_name, t2.record_type 
) as t3 
group by t3.A_BCF_name, t3.record_type; 

这里有2个嵌套查询。最深的嵌套查询就是再次显示表格，并动态添加一个额外的record_type列。下一个嵌套查询从上述嵌套查询中选择聚合在记录类型上并计算所有组。最后，外部查询聚合在A_BCF_name和record_type上，并选择每个组的最大数量。

这里是代码我写的生成样本数据来验证查询产生期望的结果：

create table dmp2g_2 (
iddmp2g_2 int unsigned primary key auto_increment, 
A_BCF_name varchar(32) not null, 
A_BTS_name varchar(32) not null, 
index (A_BCF_name), 
index (A_BTS_name) 
) engine=innodb; 

insert into dmp2g_2 (A_BCF_name, A_BTS_name) values 
('CAR.Inem', 'CAR.Inem_A'), 
('CAR.Inem', 'CAR.Inem_A'), 
('CAR.Inem', 'CAR.Inem_A'), 
('CAR.Inem', 'CAR.Inem_A'), 
('CAR.Inem', 'CAR.Inem_B'), 
('CAR.Inem', 'CAR.Inem_B'), 
('CAR.Inem', 'CAR.Inem_B'), 
('CAR.Inem', 'CAR.Inem_1'), 
('CAR.Inem', 'CAR.Inem_1'), 
('CAR.Inem', 'CAR.Inem_1'), 
('CAR.Inem', 'CAR.Inem_2'), 
('CAR.Inem', 'CAR.Inem_2'); 

而结果：

mysql> select 
    -> t3.A_BCF_name, 
    -> t3.record_type, 
    -> max(t3.record_count) as max_record_count 
    -> from 
    -> (
    -> select 
    ->  t2.A_BCF_name, 
    ->  t2.A_BTS_name, 
    ->  t2.record_type, 
    -> count(t2.record_type) record_count 
    -> from 
    -> (
    ->  select 
    ->  t1.iddmp2g_2, 
    ->  t1.A_BCF_name, 
    ->  t1.A_BTS_name, 
    ->  case when SUBSTRING_INDEX(t1.A_BTS_name,'_',-1) REGEXP '[A-Za-z]' then 'letter' else 'number' end as record_type 
    ->  from dmp2g_2 as t1 
    -> ) as t2 
    -> group by t2.A_BCF_name, t2.A_BTS_name, t2.record_type 
    -> ) as t3 
    -> group by t3.A_BCF_name, t3.record_type; 
+------------+-------------+------------------+ 
| A_BCF_name | record_type | max_record_count | 
+------------+-------------+------------------+ 
| CAR.Inem | letter  |    4 | 
| CAR.Inem | number  |    3 | 
+------------+-------------+------------------+ 
2 rows in set (0.00 sec)